Merge pull request 'Webui' (#1) from webui into main

Reviewed-on: #1

.gitignore

@@ -13,6 +13,7 @@ venv/
 *.safetensors
 models/
 weights/
+ckpts/
 
 # Audio I/O scratch
 *.wav
@@ -29,3 +30,13 @@ outputs/
 # Full-size source video — keep locally, commit only compressed derivatives
 media-files/video/*.mp4
 !media-files/video/*-480p.mp4
+
+# Compressed exports of demucs stems — keep committed for review without
+# local playback (the raw separated/*.wav stems stay local, per *.wav above)
+!demucs/*.mp3
+
+# Small public-domain demo evidence clip for the auk voice-editing demo
+!auk/*.wav
+
+# VibeVoice 4-speaker spike test clip - committed for remote review
+!vibevoice-spike/*.mp3

README.md

@@ -6,10 +6,25 @@ Audio processing demos: WhisperX, Demucs, and TTS technology proofs-of-concept.
 
 - `whisperx/` — speech-to-text with word-level alignment and diarization
 - `demucs/` — source separation (vocals/instrumentals stem splitting)
-- `tts/` — text-to-speech service demos
+- `tts/` — text-to-speech service demos (voice cloning via MOSS-TTS)
+- `auk/` — voice *editing* demos (emotion/pitch/speed/content editing on existing recordings,
+  as opposed to `tts/`'s clone-and-generate approach)
+- `webui/` — interactive pipeline studio chaining demucs -> AuK -> MOSS-TTS with previews
 
 Each subdirectory has its own README with setup and run instructions.
 
 ## Requirements
 
 Some demos are GPU-accelerated and are best run on a CUDA-capable machine (e.g. an RTX 3090).
+
+## Planned: full audio-translation demo
+
+Chain the three demos together on `media-files/` into an end-to-end pipeline:
+
+1. `demucs` — separate the source clip into vocals / instrumental stems
+2. `whisperx` — transcribe the isolated vocal stem, then machine-translate the transcript
+3. `tts` — clone the original speaker's voice (from the isolated vocal stem) and synthesize
+   the translated text in that voice, then remix with the original instrumental stem
+
+Stem separation and transcription are already working end to end (see `demucs/README.md` and
+`whisperx/README.md`); translation and voice-cloning TTS are not yet implemented.

STEVE-TEST.md

@@ -1 +0,0 @@
-Steve can edit.

auk/README.md

@@ -0,0 +1,67 @@
+# auk
+
+Voice-*editing* demos using [AuK](https://github.com/Tencent-Hunyuan/AuK) (Tencent-Hunyuan) — a
+1.5B foundation model for speech generation **and editing**. Unlike MOSS-TTS's `instruction`
+field (steers style at generation time, from a reference clip), AuK edits an *existing*
+recording directly: same speaker, same words, same duration, different delivery. It also does
+content editing (rewrite/insert/remove words), pitch/speed/volume editing, de-accenting, and
+speech enhancement/separation — all through one natural-language instruction interface.
+
+## Demo
+
+`emotion_edit_demo.py` takes the same public-domain narrator clip used by the whisperx/tts demos
+(`tts/reference/narrator_ref_en.wav`, 11.7s, isolated via demucs) and edits its emotion to
+"excited" while preserving the voice, words, and exact duration.
+
+### Setup
+
+AuK is vendored outside this repo (large third-party project, editable install), same pattern as
+MOSS-TTS:
+
+```bash
+git clone https://github.com/Tencent-Hunyuan/AuK.git /path/to/vendor/AuK
+python3 -m venv .venv && . .venv/bin/activate
+pip install -e /path/to/vendor/AuK
+
+# Two runtime deps AuK needs but doesn't declare (see Gotchas):
+pip install audioread "huggingface_hub<1.0,>=0.34.0"
+
+# Model weights (~17GB total)
+pip install -U "huggingface_hub[cli]"
+hf download tencent/AuK --local-dir ./ckpts/AuK
+hf download Qwen/Qwen2.5-Omni-3B --local-dir ./ckpts/Qwen2.5-Omni-3B
+```
+
+### Running
+
+```bash
+python3 emotion_edit_demo.py
+```
+
+Output goes to `narrator_excited.wav` (24kHz mono, ~11.7s — exact source duration preserved).
+
+## Gotchas found running this on an RTX 3090
+
+- **`qwen_omni_utils` needs `audioread`, but nothing declares it as a dependency** — same class
+  of packaging gap as demucs's missing `numpy` earlier in this repo. Install it explicitly or
+  the first real inference call fails with `ModuleNotFoundError: No module named 'audioread'`
+  (model loading itself succeeds first, so this only surfaces once generation actually starts).
+- **`pip install "huggingface_hub[cli]"` breaks `transformers`** — it pulls `huggingface_hub`
+  2.x-style 1.31.0, but `transformers` (installed as an AuK dependency) requires `<1.0`. Only
+  needed transiently to fetch the weights; downgrade back to `<1.0,>=0.34.0` before running
+  inference, or install the CLI in a separate throwaway environment.
+- **`--cpu_offload` is not optional on a 24GB card** — AuK's own documented VRAM figures
+  (24.78–25.00GiB *without* offload) exceed the RTX 3090's 24GB entirely, even dedicating the
+  whole card. Always pass `--cpu_offload` (or `cpu_offload=True` in the Python API).
+- **Real practical length limit is much shorter than it looks: ~25–30s, not 5 minutes.** AuK's
+  ComfyUI docs mention a "30-second source/reference-plus-target budget" as if it were a
+  ComfyUI-specific constraint — it isn't. A 25s clip worked fine even with `--cpu_offload`; an
+  80s clip OOM'd (tried to allocate another 8.9GB mid-attention on top of the ~17GB already in
+  use) because attention memory scales fast with sequence length. Editing a full multi-minute
+  podcast take means chunking it into ≤~25s segments and concatenating, not a single pass —
+  worth planning for prosody continuity across chunk boundaries.
+- **Default checkpoint path** baked into `auk-infer`/`AukInfer` points at wherever you cloned
+  the vendor repo, not this directory — always pass `--ckpt`/`--qwen_path` (CLI) or
+  `config_path`/`ckpt_path`/`qwen_path` (Python API) explicitly.
+- **Output is 24kHz mono**, notably lower fidelity than MOSS-TTS's 48kHz stereo output — worth
+  factoring in if AuK-edited audio needs to sit alongside MOSS-TTS output in the same pipeline.

auk/emotion_edit_demo.py

@@ -0,0 +1,47 @@
+"""
+Voice-editing demo using Tencent's AuK (https://github.com/Tencent-Hunyuan/AuK).
+
+Unlike MOSS-TTS's `instruction` field (steers style at generation time),
+AuK edits emotion on an *existing* recording directly - same speaker, same
+words, same duration, different delivery. Demonstrates on the public-domain
+narrator clip already used by the whisperx/tts demos.
+
+Run from the auk/ directory with its .venv activated.
+"""
+
+from pathlib import Path
+
+from auk.infer.infer_auk import AukInfer, save_audio
+
+HERE = Path(__file__).parent
+CONFIG = HERE / "ckpts" / "AuK" / "config.yaml"
+CKPT = HERE / "ckpts" / "AuK" / "auk_base.safetensors"
+QWEN_PATH = HERE / "ckpts" / "Qwen2.5-Omni-3B"
+INPUT_AUDIO = HERE.parent / "tts" / "reference" / "narrator_ref_en.wav"
+OUTPUT_PATH = HERE / "narrator_excited.wav"
+
+INSTRUCTION = "Change the emotion to excited."
+
+
+def main() -> None:
+    engine = AukInfer(
+        str(CONFIG), str(CKPT), qwen_path=str(QWEN_PATH), cpu_offload=True
+    )
+
+    messages = [
+        {
+            "role": "user",
+            "content": [
+                {"type": "text", "text": INSTRUCTION},
+                {"type": "audio", "audio": str(INPUT_AUDIO)},
+            ],
+        }
+    ]
+
+    audio, sr = engine.generate(messages)
+    save_audio(audio, sr, str(OUTPUT_PATH))
+    print(f"[INFO] Saved {OUTPUT_PATH}")
+
+
+if __name__ == "__main__":
+    main()

demucs/README.md

@@ -1,3 +1,30 @@
 # demucs
 
 Source separation demos using [Demucs](https://github.com/facebookresearch/demucs) — splitting audio into stems (vocals, drums, bass, other).
+
+## Demo run
+
+Ran two-stem separation (vocals vs. everything else) on an RTX 3090 against
+`media-files/audio/service-mbrs-ntscrm-01840194-01840194.m4a`:
+
+```
+python3 -m demucs --two-stems=vocals -n htdemucs \
+  -o separated \
+  ../media-files/audio/service-mbrs-ntscrm-01840194-01840194.m4a
+```
+
+Took ~12s for 11 minutes of audio. Outputs `vocals.wav` and `no_vocals.wav` (44.1kHz stereo PCM,
+duration matches source) under `separated/htdemucs/<track-name>/` — gitignored, left on the GPU
+box at `/root/projects/audio-demo/repo/demucs/separated/`.
+
+Full-length MP3 versions (128kbps, ~10MB each) are committed as `vocals.mp3` and
+`no_vocals.mp3` in this directory so both stems are reviewable directly from Gitea without
+needing local playback or GPU-box access.
+
+**Known gotcha:** demucs 4.1.0's package metadata doesn't declare `numpy` as a dependency even
+though `demucs/transformer.py` imports it directly — `pip install demucs` alone leaves the env
+broken (`ModuleNotFoundError: No module named 'numpy'` at runtime). Install `numpy` explicitly
+alongside it.
+
+The isolated `vocals.wav` is intended as the clean reference/source audio for a voice-cloning TTS
+step (see top-level README for the planned audio-translation pipeline).

mastering/master.py

@@ -0,0 +1,82 @@
+"""
+Mastering chain for generated TTS/voice clips: fixes the "tinny" spectral
+imbalance common to neural vocoder output (excess 3-6kHz presence energy,
+starved low end relative to a naturally-recorded voice - see the project's
+spectral comparison: raw recordings skew toward 200-1000Hz, generated clips
+dump ~35% of their energy into 3-6kHz) and adds the density/space that
+reads as "produced" rather than dry/synthetic.
+
+Chain order matters: EQ before compression (so the compressor reacts to
+the corrected tonal balance, not the raw tinny one), reverb after
+compression (so the room tail doesn't get squashed), limiter last.
+"""
+
+import argparse
+
+import numpy as np
+import soundfile as sf
+from pedalboard import (
+    Compressor,
+    HighpassFilter,
+    HighShelfFilter,
+    Limiter,
+    LowShelfFilter,
+    PeakFilter,
+    Pedalboard,
+    Reverb,
+)
+
+
+def build_chain(
+    warmth_db: float,
+    presence_cut_db: float,
+    air_db: float,
+    comp_ratio: float,
+    reverb_wet: float,
+) -> Pedalboard:
+    return Pedalboard([
+        HighpassFilter(cutoff_frequency_hz=60),
+        LowShelfFilter(cutoff_frequency_hz=150, gain_db=warmth_db, q=0.7),
+        PeakFilter(cutoff_frequency_hz=3800, gain_db=presence_cut_db, q=1.0),
+        HighShelfFilter(cutoff_frequency_hz=6500, gain_db=air_db, q=0.7),
+        Compressor(threshold_db=-20, ratio=comp_ratio, attack_ms=12, release_ms=120),
+        Reverb(room_size=0.08, damping=0.85, wet_level=reverb_wet, dry_level=1.0 - reverb_wet),
+        Limiter(threshold_db=-1.0, release_ms=100),
+    ])
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description="Apply a studio mastering chain to a voice clip")
+    parser.add_argument("--input", required=True)
+    parser.add_argument("--output", required=True)
+    parser.add_argument("--warmth-db", type=float, default=5.5, help="Low-shelf boost @150Hz")
+    parser.add_argument("--presence-cut-db", type=float, default=-5.0, help="Peak cut @3.8kHz (the tinny band)")
+    parser.add_argument("--air-db", type=float, default=-4.5, help="High-shelf @6.5kHz, for softness")
+    parser.add_argument("--comp-ratio", type=float, default=2.5)
+    parser.add_argument("--reverb-wet", type=float, default=0.025)
+    args = parser.parse_args()
+
+    audio, sr = sf.read(args.input, always_2d=True)
+    audio = audio.T.astype(np.float32)  # pedalboard wants (channels, samples)
+
+    board = build_chain(
+        args.warmth_db, args.presence_cut_db, args.air_db, args.comp_ratio, args.reverb_wet
+    )
+    processed = board(audio, sr)
+
+    # pedalboard's Limiter is attack/release-based, not a true brick-wall
+    # limiter - fast transients can still punch through its threshold (seen
+    # in practice: in-memory peaks hit 0dBFS despite threshold_db=-1.0,
+    # producing audible clipping distortion on playback). Hard safety
+    # ceiling here guarantees the written file never clips regardless.
+    target_peak = 10 ** (-1.0 / 20)
+    peak = np.max(np.abs(processed))
+    if peak > target_peak:
+        processed = processed * (target_peak / peak)
+
+    sf.write(args.output, processed.T, sr)
+    print(f"[INFO] Saved {args.output}")
+
+
+if __name__ == "__main__":
+    main()

sfx-stable-audio-3/generate.py

@@ -0,0 +1,32 @@
+"""
+Stable Audio 3 Small-SFX CLI: generate a sound effect/ambience clip from a
+text description. Runs on CPU by design - this checkpoint is the
+"lightweight, no GPU required" model in Stable Audio 3's lineup, so it
+never contends with ComfyUI/other models for VRAM (unlike generate_sfx.py's
+MOSS-SoundEffect, which needs the 3090 and ComfyUI's memory freed first).
+"""
+
+import argparse
+
+import torchaudio
+from stable_audio_3 import StableAudioModel
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description="Stable Audio 3 Small-SFX text-to-sound-effect")
+    parser.add_argument("--text", required=True, help="Description of the sound effect/ambience")
+    parser.add_argument("--output", required=True, help="Output wav path")
+    parser.add_argument("--duration", type=float, default=10.0, help="Clip duration in seconds (max 120)")
+    parser.add_argument("--seed", type=int, default=-1, help="Random seed, or -1 for random")
+    args = parser.parse_args()
+
+    model = StableAudioModel.from_pretrained("small-sfx", device="cpu")
+    sample_rate = model.model.sample_rate
+
+    audio = model.generate(prompt=args.text, duration=args.duration, seed=args.seed)
+    torchaudio.save(args.output, audio[0].cpu(), sample_rate)
+    print(f"[INFO] Saved {args.output}")
+
+
+if __name__ == "__main__":
+    main()

tts/generate.py

@@ -0,0 +1,109 @@
+"""
+Generic MOSS-TTS zero-shot cloning CLI: clone the voice in --reference and
+say --text in it. Used directly, and as the final stage of webui/app.py.
+"""
+
+import argparse
+import importlib.util
+
+import soundfile as sf
+import torch
+from transformers import AutoModel, AutoProcessor
+
+PRETRAINED = "OpenMOSS-Team/MOSS-TTS-Local-Transformer-v1.5"
+
+
+def resolve_attn_implementation(device: str, dtype: torch.dtype) -> str:
+    if (
+        device == "cuda"
+        and importlib.util.find_spec("flash_attn") is not None
+        and dtype in {torch.float16, torch.bfloat16}
+    ):
+        major, _ = torch.cuda.get_device_capability()
+        if major >= 8:
+            return "flash_attention_2"
+    return "sdpa" if device == "cuda" else "eager"
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description="MOSS-TTS zero-shot voice cloning")
+    parser.add_argument("--text", required=True, help="Text to synthesize")
+    parser.add_argument("--reference", required=True, help="Reference audio path")
+    parser.add_argument("--output", required=True, help="Output wav path")
+    parser.add_argument("--language", default="English", help="Language tag")
+    parser.add_argument(
+        "--instruction", default=None,
+        help='Style/delivery instruction (e.g. "Change the emotion to excited."). '
+        "Undocumented MOSS-TTS field for this checkpoint - confirmed to work for "
+        "emotion in testing, passed through as-is.",
+    )
+    parser.add_argument(
+        "--seed", type=int, default=None,
+        help="Random seed for reproducibility. Unset = nondeterministic.",
+    )
+    parser.add_argument("--audio_temperature", type=float, default=1.7)
+    parser.add_argument("--audio_top_p", type=float, default=0.8)
+    parser.add_argument("--audio_top_k", type=int, default=25)
+    parser.add_argument("--audio_repetition_penalty", type=float, default=1.0)
+    parser.add_argument("--max_new_tokens", type=int, default=4096)
+    args = parser.parse_args()
+
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    dtype = torch.bfloat16 if device == "cuda" else torch.float32
+
+    if args.seed is not None:
+        torch.manual_seed(args.seed)
+        if device == "cuda":
+            torch.cuda.manual_seed_all(args.seed)
+
+    torch.backends.cuda.enable_cudnn_sdp(False)
+    torch.backends.cuda.enable_flash_sdp(True)
+    torch.backends.cuda.enable_mem_efficient_sdp(True)
+    torch.backends.cuda.enable_math_sdp(True)
+
+    attn_implementation = resolve_attn_implementation(device, dtype)
+    print(f"[INFO] attn_implementation={attn_implementation}")
+
+    processor = AutoProcessor.from_pretrained(PRETRAINED, trust_remote_code=True)
+    processor.audio_tokenizer = processor.audio_tokenizer.to(device)
+
+    model = AutoModel.from_pretrained(
+        PRETRAINED,
+        trust_remote_code=True,
+        attn_implementation=attn_implementation,
+        dtype=dtype,
+    ).to(device)
+    model.eval()
+
+    conversation = [
+        processor.build_user_message(
+            text=args.text,
+            language=args.language,
+            reference=[args.reference],
+            instruction=args.instruction,
+        )
+    ]
+
+    with torch.no_grad():
+        batch = processor([conversation], mode="generation")
+        outputs = model.generate(
+            input_ids=batch["input_ids"].to(device),
+            attention_mask=batch["attention_mask"].to(device),
+            max_new_tokens=args.max_new_tokens,
+            do_sample=True,
+            audio_temperature=args.audio_temperature,
+            audio_top_p=args.audio_top_p,
+            audio_top_k=args.audio_top_k,
+            audio_repetition_penalty=args.audio_repetition_penalty,
+        )
+        message = processor.decode(outputs)[0]
+        audio = message.audio_codes_list[0]
+        data = audio.detach().cpu().to(torch.float32).numpy()
+        if data.ndim == 2:
+            data = data.T
+        sf.write(args.output, data, processor.model_config.sampling_rate)
+        print(f"[INFO] Saved {args.output}")
+
+
+if __name__ == "__main__":
+    main()

tts/generate_sfx.py

@@ -0,0 +1,99 @@
+"""
+MOSS-SoundEffect CLI: generate an ambient/sound-effect clip from a text
+description. Sibling of generate.py (MOSS-TTS voice cloning) - same
+backend setup and model-loading pattern, different checkpoint and a
+text-only build_user_message call (no reference audio, no cloning).
+"""
+
+import argparse
+import importlib.util
+import os
+
+# This checkpoint's weights alone use ~22.3GB of this box's ~23.56GB usable
+# VRAM, leaving generate()'s own small allocations (a vocab-sized mask
+# tensor, etc.) no room without this - confirmed via OOM otherwise. Must be
+# set before torch's CUDA allocator initializes, so it's set here rather
+# than relying on the caller's shell environment.
+os.environ.setdefault("PYTORCH_ALLOC_CONF", "expandable_segments:True")
+os.environ.setdefault("PYTORCH_CUDA_ALLOC_CONF", "expandable_segments:True")
+
+import soundfile as sf
+import torch
+from transformers import AutoModel, AutoProcessor
+
+PRETRAINED = "OpenMOSS-Team/MOSS-SoundEffect"
+
+
+def resolve_attn_implementation(device: str, dtype: torch.dtype) -> str:
+    if (
+        device == "cuda"
+        and importlib.util.find_spec("flash_attn") is not None
+        and dtype in {torch.float16, torch.bfloat16}
+    ):
+        major, _ = torch.cuda.get_device_capability()
+        if major >= 8:
+            return "flash_attention_2"
+    return "sdpa" if device == "cuda" else "eager"
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description="MOSS-SoundEffect text-to-sound-effect")
+    parser.add_argument("--text", required=True, help="Description of the sound effect/ambience")
+    parser.add_argument("--output", required=True, help="Output wav path")
+    parser.add_argument(
+        "--seed", type=int, default=None,
+        help="Random seed for reproducibility. Unset = nondeterministic.",
+    )
+    parser.add_argument("--max_new_tokens", type=int, default=4096)
+    args = parser.parse_args()
+
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    dtype = torch.bfloat16 if device == "cuda" else torch.float32
+
+    if args.seed is not None:
+        torch.manual_seed(args.seed)
+        if device == "cuda":
+            torch.cuda.manual_seed_all(args.seed)
+
+    torch.backends.cuda.enable_cudnn_sdp(False)
+    torch.backends.cuda.enable_flash_sdp(True)
+    torch.backends.cuda.enable_mem_efficient_sdp(True)
+    torch.backends.cuda.enable_math_sdp(True)
+
+    attn_implementation = resolve_attn_implementation(device, dtype)
+    print(f"[INFO] attn_implementation={attn_implementation}")
+
+    processor = AutoProcessor.from_pretrained(PRETRAINED, trust_remote_code=True)
+    processor.audio_tokenizer = processor.audio_tokenizer.to(device)
+
+    model = AutoModel.from_pretrained(
+        PRETRAINED,
+        trust_remote_code=True,
+        attn_implementation=attn_implementation,
+        dtype=dtype,
+    ).to(device)
+    model.eval()
+
+    conversation = [processor.build_user_message(ambient_sound=args.text)]
+
+    with torch.no_grad():
+        batch = processor([conversation], mode="generation")
+        # This checkpoint's generate() doesn't accept the TTS model's sampling
+        # kwargs (do_sample/audio_temperature/etc raise TypeError) - matches
+        # the official README example, which only passes these three.
+        outputs = model.generate(
+            input_ids=batch["input_ids"].to(device),
+            attention_mask=batch["attention_mask"].to(device),
+            max_new_tokens=args.max_new_tokens,
+        )
+        message = processor.decode(outputs)[0]
+        audio = message.audio_codes_list[0]
+        data = audio.detach().cpu().to(torch.float32).numpy()
+        if data.ndim == 2:
+            data = data.T
+        sf.write(args.output, data, processor.model_config.sampling_rate)
+        print(f"[INFO] Saved {args.output}")
+
+
+if __name__ == "__main__":
+    main()

vibevoice-spike/README.md

@@ -0,0 +1,52 @@
+# VibeVoice (community fork) 4-person podcast spike
+
+Standalone experiment, not integrated into the webui or the production
+pipeline. Tests whether VibeVoice (https://github.com/vibevoice-community/VibeVoice)
+can generate a 4-speaker podcast scene in one pass, cloning our four
+established character voices (Kevin, Marge, Terry, Priya) from their
+existing approved `reference_audio` in the-service-station repo.
+
+- `scripts/trailer_promo_4speaker.txt` - the test script: the real
+  "sausage rolls" trailer scene from `the-service-station/scripts/trailer-promo.md`,
+  converted from our `[Character|tags]` format to VibeVoices required
+  `Speaker N: ...` plain-text format (tags dropped - VibeVoice has no
+  per-line emotion/timing control).
+- `trailer_promo_4speaker_vibevoice-1.5b.mp3` - the generated result
+  (VibeVoice-1.5B, seed 42). 47s of audio generated in 14.5s (RTF 0.31x).
+
+Setup lives outside this repo (not committed): `.venv/` here is a fresh
+isolated venv (`pip install -e /root/projects/VibeVoice`, pinned
+`transformers==4.51.3`), pointed at the already-downloaded
+`VibeVoice-1.5B` weights under `/root/projects/ComfyUI/models/tts/VibeVoice/`.
+
+## Findings (2026-09-30)
+
+Whisper-transcript check against the script: mostly faithful, not
+word-perfect. Two small unscripted insertions ("I am here!" before Priyas
+line, "For me," before Marges reply) and one substitution ("He will." ->
+"You will."). No repetition/collapse like the AuK chain-corruption bug -
+just minor ad-libbing, not catastrophic.
+
+No audible AI disclaimer or obvious background music in this run - both
+are documented risks for this model family (see
+`the-service-station/research/vibevoice-2026-08-21.md` and
+`vibevoice-community-fork-2026-08-21.md`), so treat this one clean run as
+encouraging, not conclusive.
+
+Steves call after listening: pursue 7B (weights already downloaded, same
+venv/script, swap `--model_path`) or not worth it given the limitations
+the prior research already flagged.
+
+## 7B result (2026-09-30, same script/seed/voices)
+
+`trailer_promo_4speaker_vibevoice-7b.mp3` - VibeVoice-7B, same seed (42),
+same script, same reference voices. 34.8s of audio generated in 22.2s
+(RTF 0.64x - still faster than real-time, just less headroom than 1.5B).
+
+Whisper-transcript check: cleaner than 1.5B - neither of 1.5Bs unscripted
+insertions ("I am here!", "For me,") appeared, and "He will." came through
+exactly right this time (1.5B had drifted it to "You will."). One real
+deviation: Marges defensive line "I get one fact wrong per episode."
+came out as a question, "Can I get one fact wrong per episode?" - changes
+the delivery/meaning, not just wording. Still no audible disclaimer and
+nothing that reads as injected background music.

vibevoice-spike/scripts/trailer_promo_4speaker.txt

@@ -0,0 +1,9 @@
+Speaker 1: The Service Station. Three facts, one car park, and absolutely no reason to be in a hurry.
+Speaker 2: Except for me, obviously. I'm always in a hurry.
+Speaker 3: I had a Sinclair Spectrum in 1984, and I will find a way to bring it up.
+Speaker 1: He will.
+Speaker 4: I still don't understand why the sausage rolls are relevant, but here we are.
+Speaker 2: They're always relevant, Priya.
+Speaker 1: Every week: one underdog, one argument, and one fact Marge gets wrong.
+Speaker 2: I get one fact wrong per episode. Tops.
+Speaker 3: It's usually about football.

webui/README.md

@@ -0,0 +1,56 @@
+# webui
+
+A pipeline studio: strip a voice with **demucs**, edit it with **AuK** (emotion, pitch, speed,
+free-form instructions — with preview and undo/reset), then generate new speech in that voice
+with **MOSS-TTS**.
+
+## Why this exists
+
+Manually chaining demucs → AuK → MOSS-TTS across three separate venvs (via SSH one-off scripts)
+worked but made it slow to experiment — especially with AuK's pitch/emotion controls, where you
+really want to preview a result and back out if it's wrong (see the pitch-overshoot findings in
+`auk/README.md`) rather than committing to a specific edit blind.
+
+## Architecture
+
+A single lightweight Gradio app (`app.py`) that **shells out** to each tool's own isolated venv
+via subprocess, rather than merging their dependencies into one environment — MOSS-TTS pins
+`transformers==5.0.0`, AuK needs `4.57.6`; they can't coexist in one venv. This app's own venv
+only needs `gradio` itself.
+
+Each stage's output becomes the next stage's input by filesystem path; a `history` list backs
+the undo/reset buttons in the voice editor.
+
+## Setup
+
+```bash
+python3 -m venv .venv && . .venv/bin/activate
+pip install gradio
+```
+
+Assumes `demucs/`, `auk/`, and `tts/` (siblings in this repo) already have their own working
+venvs and, for `auk/`, downloaded model weights — see their READMEs.
+
+## Running
+
+```bash
+python3 app.py
+```
+
+Serves on `0.0.0.0:7860`. The app automatically calls ComfyUI's `/free` endpoint (if its queue
+is empty) before each heavy step, to reduce GPU contention on a shared box — see the DNS/routing
+and GPU-contention notes earlier in this project's history for why that matters here.
+
+## Known limitations (inherited from the underlying tools)
+
+- **AuK's real usable clip length is ~25-30s**, not several minutes — feeding it a long recording
+  will OOM even with `--cpu_offload` (see `auk/README.md`). This app doesn't chunk long inputs
+  automatically; keep reference clips short.
+- **AuK's pitch-editing instruction-following is unreliable** — even "1 semitone" has measured
+  out at ~4+ semitones of actual shift in testing. The UI surfaces a warning; preview before
+  trusting a pitch edit.
+- Each stage is a blocking subprocess call (model load + inference each time) — the UI will
+  appear to hang for anywhere from ~10s to a couple of minutes per click, with no progress bar
+  yet.
+- No queueing/locking across browser tabs or concurrent users — this is a single-GPU box, so two
+  simultaneous generations will contend for VRAM.

whisperx/README.md

@@ -1,19 +0,0 @@
-# whisperx
-
-Speech-to-text demos using [WhisperX](https://github.com/m-bain/whisperX) — fast transcription with word-level timestamps and speaker diarization.
-
-## Demo run
-
-Ran on an RTX 3090 against `media-files/audio/service-mbrs-ntscrm-01840194-01840194.m4a`
-(audio track extracted from the sample video in `media-files/video/`):
-
-```
-whisperx media-files/audio/service-mbrs-ntscrm-01840194-01840194.m4a \
-  --model medium \
-  --output_dir whisperx/transcripts \
-  --output_format all \
-  --compute_type float16
-```
-
-Outputs (json/srt/tsv/txt/vtt) are in `whisperx/transcripts/`. Language auto-detected as English;
-sentence-level timestamps aligned via wav2vec2.

whisperx/transcripts/service-mbrs-ntscrm-01840194-01840194.json

@@ -1 +0,0 @@
-{"segments": [{"start": 0.847, "end": 3.369, "text": " from the Library of Congress in Washington, D.C.", "words": [{"word": "from", "start": 0.847, "end": 1.007, "score": 0.795}, {"word": "the", "start": 1.027, "end": 1.107, "score": 0.832}, {"word": "Library", "start": 1.147, "end": 1.647, "score": 0.894}, {"word": "of", "start": 1.668, "end": 1.748, "score": 0.5}, {"word": "Congress", "start": 1.808, "end": 2.268, "score": 0.771}, {"word": "in", "start": 2.448, "end": 2.528, "score": 0.832}, {"word": "Washington,", "start": 2.568, "end": 3.088, "score": 0.869}, {"word": "D.C.", "start": 3.128, "end": 3.369, "score": 0.953}], "avg_logprob": -0.08811598557692307}, {"start": 35.875, "end": 47.583, "text": " We shall show you in this film typical examples of children who have been well brought up, the kind who grow up to be strong, healthy, and happy like these young folks.", "words": [{"word": "We", "start": 35.875, "end": 35.975, "score": 0.999}, {"word": "shall", "start": 36.016, "end": 36.216, "score": 0.867}, {"word": "show", "start": 36.276, "end": 36.516, "score": 0.9}, {"word": "you", "start": 36.576, "end": 36.756, "score": 0.863}, {"word": "in", "start": 36.916, "end": 36.996, "score": 0.842}, {"word": "this", "start": 37.016, "end": 37.196, "score": 0.881}, {"word": "film", "start": 37.336, "end": 37.697, "score": 0.696}, {"word": "typical", "start": 37.717, "end": 38.377, "score": 0.661}, {"word": "examples", "start": 38.437, "end": 38.917, "score": 0.572}, {"word": "of", "start": 38.977, "end": 39.057, "score": 0.46}, {"word": "children", "start": 39.138, "end": 39.578, "score": 0.933}, {"word": "who", "start": 39.618, "end": 39.718, "score": 0.904}, {"word": "have", "start": 39.738, "end": 39.858, "score": 0.783}, {"word": "been", "start": 39.898, "end": 40.058, "score": 0.751}, {"word": "well", "start": 40.118, "end": 40.358, "score": 0.852}, {"word": "brought", "start": 40.398, "end": 40.658, "score": 0.853}, {"word": "up,", "start": 40.719, "end": 40.819, "score": 0.938}, {"word": "the", "start": 41.499, "end": 41.599, "score": 0.665}, {"word": "kind", "start": 41.639, "end": 41.979, "score": 0.835}, {"word": "who", "start": 41.999, "end": 42.079, "score": 0.147}, {"word": "grow", "start": 42.139, "end": 42.38, "score": 0.699}, {"word": "up", "start": 42.42, "end": 42.5, "score": 0.956}, {"word": "to", "start": 42.52, "end": 42.7, "score": 0.404}, {"word": "be", "start": 42.72, "end": 42.88, "score": 0.83}, {"word": "strong,", "start": 42.9, "end": 43.4, "score": 0.659}, {"word": "healthy,", "start": 43.881, "end": 44.321, "score": 0.857}, {"word": "and", "start": 44.441, "end": 44.521, "score": 0.982}, {"word": "happy", "start": 44.581, "end": 44.941, "score": 0.986}, {"word": "like", "start": 45.382, "end": 45.582, "score": 0.888}, {"word": "these", "start": 45.662, "end": 45.942, "score": 0.704}, {"word": "young", "start": 46.002, "end": 46.262, "score": 0.663}, {"word": "folks.", "start": 46.302, "end": 47.583, "score": 0.617}], "avg_logprob": -0.06059027806752258}, {"start": 47.603, "end": 61.552, "text": "Sturdy growth, joyousness, contentment in life, and vigorous health for the children of today will build men and women of tomorrow with strength of body, nerve, and mind, and thus a better chance to get the most out of life.", "words": [{"word": "Sturdy", "start": 47.603, "end": 47.963, "score": 0.622}, {"word": "growth,", "start": 48.043, "end": 48.584, "score": 0.655}, {"word": "joyousness,", "start": 48.744, "end": 49.404, "score": 0.463}, {"word": "contentment", "start": 49.584, "end": 50.185, "score": 0.867}, {"word": "in", "start": 50.225, "end": 50.305, "score": 0.93}, {"word": "life,", "start": 50.365, "end": 50.645, "score": 0.935}, {"word": "and", "start": 50.725, "end": 50.825, "score": 0.805}, {"word": "vigorous", "start": 50.905, "end": 51.285, "score": 0.552}, {"word": "health", "start": 51.325, "end": 51.626, "score": 0.541}, {"word": "for", "start": 51.766, "end": 51.886, "score": 0.57}, {"word": "the", "start": 51.906, "end": 51.966, "score": 0.992}, {"word": "children", "start": 52.026, "end": 52.306, "score": 0.891}, {"word": "of", "start": 52.346, "end": 52.386, "score": 0.997}, {"word": "today", "start": 52.446, "end": 52.826, "score": 0.762}, {"word": "will", "start": 52.846, "end": 53.587, "score": 0.406}, {"word": "build", "start": 53.607, "end": 53.927, "score": 0.762}, {"word": "men", "start": 54.047, "end": 54.247, "score": 0.916}, {"word": "and", "start": 54.287, "end": 54.367, "score": 0.847}, {"word": "women", "start": 54.427, "end": 54.667, "score": 0.873}, {"word": "of", "start": 54.688, "end": 54.748, "score": 0.592}, {"word": "tomorrow", "start": 54.808, "end": 55.308, "score": 0.833}, {"word": "with", "start": 55.428, "end": 55.568, "score": 0.986}, {"word": "strength", "start": 55.628, "end": 55.968, "score": 0.814}, {"word": "of", "start": 56.008, "end": 56.068, "score": 0.748}, {"word": "body,", "start": 56.168, "end": 56.509, "score": 0.993}, {"Line truncated

whisperx/transcripts/service-mbrs-ntscrm-01840194-01840194.srt

@@ -1,380 +0,0 @@
-1
-00:00:00,847 --> 00:00:03,369
-from the Library of Congress in Washington, D.C.
-
-2
-00:00:35,875 --> 00:00:47,583
-We shall show you in this film typical examples of children who have been well brought up, the kind who grow up to be strong, healthy, and happy like these young folks.
-
-3
-00:00:47,603 --> 00:01:01,552
-Sturdy growth, joyousness, contentment in life, and vigorous health for the children of today will build men and women of tomorrow with strength of body, nerve, and mind, and thus a better chance to get the most out of life.
-
-4
-00:01:14,263 --> 00:01:22,830
-These healthy children are not all from well-to-do homes, but from babyhood have had the right food, sunshine, companionship, and intelligent care.
-
-5
-00:01:23,510 --> 00:01:25,572
-Your child can have all of this too.
-
-6
-00:01:25,972 --> 00:01:28,454
-Do you really see your child when you look at him?
-
-7
-00:01:29,875 --> 00:01:31,717
-Look at David, age nine.
-
-8
-00:01:32,137 --> 00:01:33,358
-How straight he stands.
-
-9
-00:01:34,122 --> 00:01:40,427
-Note his glossy hair, his good color, his broad chest with room for his lungs to breathe.
-
-10
-00:01:41,448 --> 00:01:50,715
-David has none of those enlargements of bones of shoulder, chest, ribs, knees, and ankles due to poor bone growth.
-
-11
-00:01:51,356 --> 00:01:52,917
-He has well-shaped knees.
-
-12
-00:01:53,898 --> 00:01:55,419
-He stands on straight legs.
-
-13
-00:01:56,080 --> 00:02:03,686
-He has beautiful feet with small ankle bones and fine arches, straight toes, not kinked under like some of ours.
-
-14
-00:02:04,220 --> 00:02:12,004
-He has good muscles and uses them well, and is well because his parents have given him a chance from infancy.
-
-15
-00:02:18,348 --> 00:02:21,670
-Like David, little Milo has been breastfed since birth.
-
-16
-00:02:22,570 --> 00:02:24,911
-Breast food is baby's best food.
-
-17
-00:02:26,312 --> 00:02:31,215
-Milo, three months getting his daily exercise, already he raises himself up.
-
-18
-00:02:32,894 --> 00:02:34,795
-Helen getting her sun bath.
-
-19
-00:02:35,395 --> 00:02:37,496
-At three months, her back is well shaped.
-
-20
-00:02:37,996 --> 00:02:39,177
-She holds her head erect.
-
-21
-00:02:39,757 --> 00:02:43,139
-The straight lines down her side show good development.
-
-22
-00:02:45,920 --> 00:02:48,341
-When baby is ready, he'll pull himself up.
-
-23
-00:02:49,382 --> 00:02:52,023
-Seven-month-old Bill delights in doing this.
-
-24
-00:02:52,883 --> 00:02:56,785
-Don't jump your baby up and down until he does it for himself.
-
-25
-00:02:58,618 --> 00:03:03,361
-At 10 months, small bone, beautifully developed genes gets her exercise.
-
-26
-00:03:03,802 --> 00:03:06,063
-Give babies plenty of space to creep.
-
-27
-00:03:07,604 --> 00:03:11,887
-At 13 months, Ann thrills her parents by taking her first step alone.
-
-28
-00:03:12,507 --> 00:03:16,950
-Look at her broad chest and fairly flat tummy, though she has just had her dinner.
-
-29
-00:03:17,951 --> 00:03:19,872
-Ann loves to stand holding to the buggy.
-
-30
-00:03:20,333 --> 00:03:24,616
-Note her nice straight back, her straight lines down her side showing good development.
-
-31
-00:03:25,196 --> 00:03:26,677
-She has a well-shaped head,
-
-32
-00:03:27,150 --> 00:03:29,692
-good color and abundant glossy curls.
-
-33
-00:03:37,098 --> 00:03:39,820
-Jimmy, the same age as Ann, is stockier built.
-
-34
-00:03:40,041 --> 00:03:41,602
-His bones and muscles are heavier.
-
-35
-00:03:41,962 --> 00:03:45,265
-He walked sooner, partly because he had more chance to creep.
-
-36
-00:03:47,867 --> 00:03:52,751
-All these preschool children have been well fed since birth and have good appetites.
-
-37
-00:03:56,623 --> 00:03:59,224
-Agnes and Carl both two and a half years.
-
-38
-00:03:59,764 --> 00:04:04,566
-Carl has a well-shaped head, glossy hair, and very good color.
-
-39
-00:04:05,126 --> 00:04:10,568
-As he watches his dandelion sail away presently, see what fine teeth he has.
-
-40
-00:04:11,068 --> 00:04:18,791
-How straight and even they are in his broad, well-developed jaw, which gives plenty of room for his second teeth.
-
-41
-00:04:22,051 --> 00:04:27,192
-Joyous fairy like Patsy with Carly's beautifully developed child who rejoices in the activity.
-
-42
-00:04:27,652 --> 00:04:29,913
-Patsy at four years knows her foods.
-
-43
-00:04:30,313 --> 00:04:36,835
-She said to her aunt, I mustn't drink cocoa, tea, coffee, cola, nothing but milk.
-
-44
-00:04:42,676 --> 00:04:46,237
-Patsy is going to serve milk to her little friends.
-
-45
-00:04:47,317 --> 00:04:51,258
-Patsy and dark haired Marilyn are both four years old.
-
-46
-00:05:02,408 --> 00:05:07,652
-Bobby Six and his sister Irene Four have well-grown bones and strong muscles.
-
-47
-00:05:08,452 --> 00:05:10,153
-What good control they have of them.
-
-48
-00:05:10,854 --> 00:05:15,737
-Bobby's perfectly developed chest and ribs give plenty of room for breathing.
-
-49
-00:05:18,359 --> 00:05:24,744
-Bobby also, you will note, has well-spaced, even teeth in a well-developed jaw.
-
-50
-00:05:26,980 --> 00:05:30,301
-Note these happy seven to 11-year-old children.
-
-51
-00:05:31,061 --> 00:05:37,484
-Barbara, standing erect in the foreground calling to Joan, is well-developed, tall, with broad shoulders.
-
-52
-00:05:38,344 --> 00:05:41,305
-Barbara and Joan are both seven years old.
-
-53
-00:05:42,125 --> 00:05:45,967
-Joan, too, is well-developed, but short and stalky.
-
-54
-00:05:46,847 --> 00:05:49,828
-This older group shows differences in size and build.
-
-55
-00:05:50,128 --> 00:05:54,110
-Harriet, the little one in the center, is 11 years old, and Marilyn, the largest, 13.
-
-56
-00:05:58,040 --> 00:06:01,062
-This is David, nine years old, and Don, twelve.
-
-57
-00:06:01,442 --> 00:06:03,183
-Don is a thin, wiry type.
-
-58
-00:06:03,723 --> 00:06:05,864
-David, whom we have met before, is heavy.
-
-59
-00:06:12,248 --> 00:06:25,975
-Jean, eleven years old, shows beautiful development and excellent control of her muscles, and is a lovely-looking creature, because from infancy, Jean has had everything that helps to make a beautiful human being.
-
-60
-00:06:32,601 --> 00:06:37,683
-Alan and his sister Betty illustrate the racehorse type with long, slender bones and muscles.
-
-61
-00:06:38,243 --> 00:06:43,006
-Both are still growing rapidly and need plenty of the right kind of food and rest.
-
-62
-00:06:44,146 --> 00:06:49,849
-Alan shows excellent control of his long, slender muscles, carries himself well and breathes well.
-
-63
-00:06:50,509 --> 00:06:55,272
-Alan's charming smile as he chins shows vitality and joy in living.
-
-64
-00:06:56,586 --> 00:07:01,171
-These children, helping mother pick vegetables, get many kinds of nourishment.
-
-65
-00:07:01,731 --> 00:07:05,775
-Fresh air, sunshine, vegetables, eggs, too.
-
-66
-00:07:07,136 --> 00:07:11,761
-Now, note the following necessary foods.
-
-67
-00:07:14,003 --> 00:07:16,866
-Milk, the best bone building food for all ages.
-
-68
-00:07:18,167 --> 00:07:20,610
-Oranges, an excellent source of vitamin C.
-
-69
-00:07:26,008 --> 00:07:29,830
-Also cabbage and tomatoes, cabbage usually not for two years.
-
-70
-00:07:30,911 --> 00:07:34,153
-Potato is a cheap source of iron, vitamin C, and energy.
-
-71
-00:07:34,953 --> 00:07:37,995
-Green and yellow vegetables, strained of course for baby.
-
-72
-00:07:43,559 --> 00:07:46,520
-Dried beans for blood and muscle, not for young children.
-
-73
-00:07:48,382 --> 00:07:50,743
-This cut of beef costs little.
-
-74
-00:07:51,373 --> 00:07:55,977
-Pig's liver at 12 cents a pound has all this cut has, plus all the vitamins.
-
-75
-00:07:56,417 --> 00:07:57,758
-Strained liver for baby.
-
-76
-00:07:58,639 --> 00:08:00,901
-Dried fruits strained, of course, for baby.
-
-77
-00:08:04,945 --> 00:08:05,985
-Fresh fruits too.
-
-78
-00:08:08,868 --> 00:08:10,790
-Bread whole-grained after two years.
-
-79
-00:08:16,414 --> 00:08:19,497
-Whole-grained cereal like rolled oats with milk is fine for breakfast.
-
-80
-00:08:20,341 --> 00:08:21,261
-Strain for baby.
-
-81
-00:08:23,603 --> 00:08:25,884
-Eggs, one a day for each child.
-
-82
-00:08:31,806 --> 00:08:35,168
-Salmon or other fat sea fish, a cheap form of protein food.
-
-83
-00:08:35,648 --> 00:08:40,050
-The strained oil can take the place of some of your cod liver oil, even for baby.
-
-84
-00:08:41,571 --> 00:08:46,853
-Bacon, an appetizing energy food, especially for older growing boys and girls.
-
-85
-00:08:50,540 --> 00:08:52,181
-should give more than just energy.
-
-86
-00:08:52,581 --> 00:08:55,363
-Molasses and brown sugar give some iron also.
-
-87
-00:08:55,844 --> 00:08:56,964
-White sugar does not.
-
-88
-00:09:00,046 --> 00:09:03,629
-Well-nurtured children become healthy and happy grown-ups.
-
-89
-00:09:04,429 --> 00:09:20,180
-An abundant and properly balanced diet, based on the foods just seen, plus plenty of sunlight and play and a due portion of the milk of human kindness, makes children grow into healthy, well-balanced, efficient citizens.
-
-90
-00:09:31,286 --> 00:09:35,067
-All the youngsters seen in these pictures have had such care.
-
-91
-00:09:45,970 --> 00:09:50,992
-We would see to it that the boys and girls in our homes and on our playgrounds
-
-92
-00:09:51,523 --> 00:10:06,536
-have the inexpensive yet priceless advantages these youngsters have had and are thus well fitted to take their places in the ranks of those who do the everyday work of the world, the work on which our national welfare depends.
-
-93
-00:10:19,102 --> 00:10:20,624
-Oh, my God.
-
-94
-00:10:52,411 --> 00:10:55,504
-This has been a presentation of the Library of Congress.
-
-95
-00:10:56,106 --> 00:10:57,953
-Visit us at loc.gov.
-

whisperx/transcripts/service-mbrs-ntscrm-01840194-01840194.tsv

@@ -1,96 +0,0 @@
-start	end	text
-847	3369	from the Library of Congress in Washington, D.C.
-35875	47583	We shall show you in this film typical examples of children who have been well brought up, the kind who grow up to be strong, healthy, and happy like these young folks.
-47603	61552	Sturdy growth, joyousness, contentment in life, and vigorous health for the children of today will build men and women of tomorrow with strength of body, nerve, and mind, and thus a better chance to get the most out of life.
-74263	82830	These healthy children are not all from well-to-do homes, but from babyhood have had the right food, sunshine, companionship, and intelligent care.
-83510	85572	Your child can have all of this too.
-85972	88454	Do you really see your child when you look at him?
-89875	91717	Look at David, age nine.
-92137	93358	How straight he stands.
-94122	100427	Note his glossy hair, his good color, his broad chest with room for his lungs to breathe.
-101448	110715	David has none of those enlargements of bones of shoulder, chest, ribs, knees, and ankles due to poor bone growth.
-111356	112917	He has well-shaped knees.
-113898	115419	He stands on straight legs.
-116080	123686	He has beautiful feet with small ankle bones and fine arches, straight toes, not kinked under like some of ours.
-124220	132004	He has good muscles and uses them well, and is well because his parents have given him a chance from infancy.
-138348	141670	Like David, little Milo has been breastfed since birth.
-142570	144911	Breast food is baby's best food.
-146312	151215	Milo, three months getting his daily exercise, already he raises himself up.
-152894	154795	Helen getting her sun bath.
-155395	157496	At three months, her back is well shaped.
-157996	159177	She holds her head erect.
-159757	163139	The straight lines down her side show good development.
-165920	168341	When baby is ready, he'll pull himself up.
-169382	172023	Seven-month-old Bill delights in doing this.
-172883	176785	Don't jump your baby up and down until he does it for himself.
-178618	183361	At 10 months, small bone, beautifully developed genes gets her exercise.
-183802	186063	Give babies plenty of space to creep.
-187604	191887	At 13 months, Ann thrills her parents by taking her first step alone.
-192507	196950	Look at her broad chest and fairly flat tummy, though she has just had her dinner.
-197951	199872	Ann loves to stand holding to the buggy.
-200333	204616	Note her nice straight back, her straight lines down her side showing good development.
-205196	206677	She has a well-shaped head,
-207150	209692	good color and abundant glossy curls.
-217098	219820	Jimmy, the same age as Ann, is stockier built.
-220041	221602	His bones and muscles are heavier.
-221962	225265	He walked sooner, partly because he had more chance to creep.
-227867	232751	All these preschool children have been well fed since birth and have good appetites.
-236623	239224	Agnes and Carl both two and a half years.
-239764	244566	Carl has a well-shaped head, glossy hair, and very good color.
-245126	250568	As he watches his dandelion sail away presently, see what fine teeth he has.
-251068	258791	How straight and even they are in his broad, well-developed jaw, which gives plenty of room for his second teeth.
-262051	267192	Joyous fairy like Patsy with Carly's beautifully developed child who rejoices in the activity.
-267652	269913	Patsy at four years knows her foods.
-270313	276835	She said to her aunt, I mustn't drink cocoa, tea, coffee, cola, nothing but milk.
-282676	286237	Patsy is going to serve milk to her little friends.
-287317	291258	Patsy and dark haired Marilyn are both four years old.
-302408	307652	Bobby Six and his sister Irene Four have well-grown bones and strong muscles.
-308452	310153	What good control they have of them.
-310854	315737	Bobby's perfectly developed chest and ribs give plenty of room for breathing.
-318359	324744	Bobby also, you will note, has well-spaced, even teeth in a well-developed jaw.
-326980	330301	Note these happy seven to 11-year-old children.
-331061	337484	Barbara, standing erect in the foreground calling to Joan, is well-developed, tall, with broad shoulders.
-338344	341305	Barbara and Joan are both seven years old.
-342125	345967	Joan, too, is well-developed, but short and stalky.
-346847	349828	This older group shows differences in size and build.
-350128	354110	Harriet, the little one in the center, is 11 years old, and Marilyn, the largest, 13.
-358040	361062	This is David, nine years old, and Don, twelve.
-361442	363183	Don is a thin, wiry type.
-363723	365864	David, whom we have met before, is heavy.
-372248	385975	Jean, eleven years old, shows beautiful development and excellent control of her muscles, and is a lovely-looking creature, because from infancy, Jean has had everything that helps to make a beautiful human being.
-392601	397683	Alan and his sister Betty illustrate the racehorse type with long, slender bones and muscles.
-398243	403006	Both are still growing rapidly and need plenty of the right kind of food and rest.
-404146	409849	Alan shows excellent control of his long, slender muscles, carries himself well and breathes well.
-410509	415272	Alan's charming smile as he chins shows vitality and joy in living.
-416586	421171	These children, helping mother pick vegetables, get many kinds of nourishment.
-421731	425775	Fresh air, sunshine, vegetables, eggs, too.
-427136	431761	Now, note the following necessary foods.
-434003	436866	Milk, the best bone building food for all ages.
-438167	440610	Oranges, an excellent source of vitamin C.
-446008	449830	Also cabbage and tomatoes, cabbage usually not for two years.
-450911	454153	Potato is a cheap source of iron, vitamin C, and energy.
-454953	457995	Green and yellow vegetables, strained of course for baby.
-463559	466520	Dried beans for blood and muscle, not for young children.
-468382	470743	This cut of beef costs little.
-471373	475977	Pig's liver at 12 cents a pound has all this cut has, plus all the vitamins.
-476417	477758	Strained liver for baby.
-478639	480901	Dried fruits strained, of course, for baby.
-484945	485985	Fresh fruits too.
-488868	490790	Bread whole-grained after two years.
-496414	499497	Whole-grained cereal like rolled oats with milk is fine for breakfast.
-500341	501261	Strain for baby.
-503603	505884	Eggs, one a day for each child.
-511806	515168	Salmon or other fat sea fish, a cheap form of protein food.
-515648	520050	The strained oil can take the place of some of your cod liver oil, even for baby.
-521571	526853	Bacon, an appetizing energy food, especially for older growing boys and girls.
-530540	532181	should give more than just energy.
-532581	535363	Molasses and brown sugar give some iron also.
-535844	536964	White sugar does not.
-540046	543629	Well-nurtured children become healthy and happy grown-ups.
-544429	560180	An abundant and properly balanced diet, based on the foods just seen, plus plenty of sunlight and play and a due portion of the milk of human kindness, makes children grow into healthy, well-balanced, efficient citizens.
-571286	575067	All the youngsters seen in these pictures have had such care.
-585970	590992	We would see to it that the boys and girls in our homes and on our playgrounds
-591523	606536	have the inexpensive yet priceless advantages these youngsters have had and are thus well fitted to take their places in the ranks of those who do the everyday work of the world, the work on which our national welfare depends.
-619102	620624	Oh, my God.
-652411	655504	This has been a presentation of the Library of Congress.
-656106	657953	Visit us at loc.gov.

whisperx/transcripts/service-mbrs-ntscrm-01840194-01840194.txt

@@ -1,95 +0,0 @@
-from the Library of Congress in Washington, D.C.
-We shall show you in this film typical examples of children who have been well brought up, the kind who grow up to be strong, healthy, and happy like these young folks.
-Sturdy growth, joyousness, contentment in life, and vigorous health for the children of today will build men and women of tomorrow with strength of body, nerve, and mind, and thus a better chance to get the most out of life.
-These healthy children are not all from well-to-do homes, but from babyhood have had the right food, sunshine, companionship, and intelligent care.
-Your child can have all of this too.
-Do you really see your child when you look at him?
-Look at David, age nine.
-How straight he stands.
-Note his glossy hair, his good color, his broad chest with room for his lungs to breathe.
-David has none of those enlargements of bones of shoulder, chest, ribs, knees, and ankles due to poor bone growth.
-He has well-shaped knees.
-He stands on straight legs.
-He has beautiful feet with small ankle bones and fine arches, straight toes, not kinked under like some of ours.
-He has good muscles and uses them well, and is well because his parents have given him a chance from infancy.
-Like David, little Milo has been breastfed since birth.
-Breast food is baby's best food.
-Milo, three months getting his daily exercise, already he raises himself up.
-Helen getting her sun bath.
-At three months, her back is well shaped.
-She holds her head erect.
-The straight lines down her side show good development.
-When baby is ready, he'll pull himself up.
-Seven-month-old Bill delights in doing this.
-Don't jump your baby up and down until he does it for himself.
-At 10 months, small bone, beautifully developed genes gets her exercise.
-Give babies plenty of space to creep.
-At 13 months, Ann thrills her parents by taking her first step alone.
-Look at her broad chest and fairly flat tummy, though she has just had her dinner.
-Ann loves to stand holding to the buggy.
-Note her nice straight back, her straight lines down her side showing good development.
-She has a well-shaped head,
-good color and abundant glossy curls.
-Jimmy, the same age as Ann, is stockier built.
-His bones and muscles are heavier.
-He walked sooner, partly because he had more chance to creep.
-All these preschool children have been well fed since birth and have good appetites.
-Agnes and Carl both two and a half years.
-Carl has a well-shaped head, glossy hair, and very good color.
-As he watches his dandelion sail away presently, see what fine teeth he has.
-How straight and even they are in his broad, well-developed jaw, which gives plenty of room for his second teeth.
-Joyous fairy like Patsy with Carly's beautifully developed child who rejoices in the activity.
-Patsy at four years knows her foods.
-She said to her aunt, I mustn't drink cocoa, tea, coffee, cola, nothing but milk.
-Patsy is going to serve milk to her little friends.
-Patsy and dark haired Marilyn are both four years old.
-Bobby Six and his sister Irene Four have well-grown bones and strong muscles.
-What good control they have of them.
-Bobby's perfectly developed chest and ribs give plenty of room for breathing.
-Bobby also, you will note, has well-spaced, even teeth in a well-developed jaw.
-Note these happy seven to 11-year-old children.
-Barbara, standing erect in the foreground calling to Joan, is well-developed, tall, with broad shoulders.
-Barbara and Joan are both seven years old.
-Joan, too, is well-developed, but short and stalky.
-This older group shows differences in size and build.
-Harriet, the little one in the center, is 11 years old, and Marilyn, the largest, 13.
-This is David, nine years old, and Don, twelve.
-Don is a thin, wiry type.
-David, whom we have met before, is heavy.
-Jean, eleven years old, shows beautiful development and excellent control of her muscles, and is a lovely-looking creature, because from infancy, Jean has had everything that helps to make a beautiful human being.
-Alan and his sister Betty illustrate the racehorse type with long, slender bones and muscles.
-Both are still growing rapidly and need plenty of the right kind of food and rest.
-Alan shows excellent control of his long, slender muscles, carries himself well and breathes well.
-Alan's charming smile as he chins shows vitality and joy in living.
-These children, helping mother pick vegetables, get many kinds of nourishment.
-Fresh air, sunshine, vegetables, eggs, too.
-Now, note the following necessary foods.
-Milk, the best bone building food for all ages.
-Oranges, an excellent source of vitamin C.
-Also cabbage and tomatoes, cabbage usually not for two years.
-Potato is a cheap source of iron, vitamin C, and energy.
-Green and yellow vegetables, strained of course for baby.
-Dried beans for blood and muscle, not for young children.
-This cut of beef costs little.
-Pig's liver at 12 cents a pound has all this cut has, plus all the vitamins.
-Strained liver for baby.
-Dried fruits strained, of course, for baby.
-Fresh fruits too.
-Bread whole-grained after two years.
-Whole-grained cereal like rolled oats with milk is fine for breakfast.
-Strain for baby.
-Eggs, one a day for each child.
-Salmon or other fat sea fish, a cheap form of protein food.
-The strained oil can take the place of some of your cod liver oil, even for baby.
-Bacon, an appetizing energy food, especially for older growing boys and girls.
-should give more than just energy.
-Molasses and brown sugar give some iron also.
-White sugar does not.
-Well-nurtured children become healthy and happy grown-ups.
-An abundant and properly balanced diet, based on the foods just seen, plus plenty of sunlight and play and a due portion of the milk of human kindness, makes children grow into healthy, well-balanced, efficient citizens.
-All the youngsters seen in these pictures have had such care.
-We would see to it that the boys and girls in our homes and on our playgrounds
-have the inexpensive yet priceless advantages these youngsters have had and are thus well fitted to take their places in the ranks of those who do the everyday work of the world, the work on which our national welfare depends.
-Oh, my God.
-This has been a presentation of the Library of Congress.
-Visit us at loc.gov.

whisperx/transcripts/service-mbrs-ntscrm-01840194-01840194.vtt

@@ -1,287 +0,0 @@
-WEBVTT
-
-00:00.847 --> 00:03.369
-from the Library of Congress in Washington, D.C.
-
-00:35.875 --> 00:47.583
-We shall show you in this film typical examples of children who have been well brought up, the kind who grow up to be strong, healthy, and happy like these young folks.
-
-00:47.603 --> 01:01.552
-Sturdy growth, joyousness, contentment in life, and vigorous health for the children of today will build men and women of tomorrow with strength of body, nerve, and mind, and thus a better chance to get the most out of life.
-
-01:14.263 --> 01:22.830
-These healthy children are not all from well-to-do homes, but from babyhood have had the right food, sunshine, companionship, and intelligent care.
-
-01:23.510 --> 01:25.572
-Your child can have all of this too.
-
-01:25.972 --> 01:28.454
-Do you really see your child when you look at him?
-
-01:29.875 --> 01:31.717
-Look at David, age nine.
-
-01:32.137 --> 01:33.358
-How straight he stands.
-
-01:34.122 --> 01:40.427
-Note his glossy hair, his good color, his broad chest with room for his lungs to breathe.
-
-01:41.448 --> 01:50.715
-David has none of those enlargements of bones of shoulder, chest, ribs, knees, and ankles due to poor bone growth.
-
-01:51.356 --> 01:52.917
-He has well-shaped knees.
-
-01:53.898 --> 01:55.419
-He stands on straight legs.
-
-01:56.080 --> 02:03.686
-He has beautiful feet with small ankle bones and fine arches, straight toes, not kinked under like some of ours.
-
-02:04.220 --> 02:12.004
-He has good muscles and uses them well, and is well because his parents have given him a chance from infancy.
-
-02:18.348 --> 02:21.670
-Like David, little Milo has been breastfed since birth.
-
-02:22.570 --> 02:24.911
-Breast food is baby's best food.
-
-02:26.312 --> 02:31.215
-Milo, three months getting his daily exercise, already he raises himself up.
-
-02:32.894 --> 02:34.795
-Helen getting her sun bath.
-
-02:35.395 --> 02:37.496
-At three months, her back is well shaped.
-
-02:37.996 --> 02:39.177
-She holds her head erect.
-
-02:39.757 --> 02:43.139
-The straight lines down her side show good development.
-
-02:45.920 --> 02:48.341
-When baby is ready, he'll pull himself up.
-
-02:49.382 --> 02:52.023
-Seven-month-old Bill delights in doing this.
-
-02:52.883 --> 02:56.785
-Don't jump your baby up and down until he does it for himself.
-
-02:58.618 --> 03:03.361
-At 10 months, small bone, beautifully developed genes gets her exercise.
-
-03:03.802 --> 03:06.063
-Give babies plenty of space to creep.
-
-03:07.604 --> 03:11.887
-At 13 months, Ann thrills her parents by taking her first step alone.
-
-03:12.507 --> 03:16.950
-Look at her broad chest and fairly flat tummy, though she has just had her dinner.
-
-03:17.951 --> 03:19.872
-Ann loves to stand holding to the buggy.
-
-03:20.333 --> 03:24.616
-Note her nice straight back, her straight lines down her side showing good development.
-
-03:25.196 --> 03:26.677
-She has a well-shaped head,
-
-03:27.150 --> 03:29.692
-good color and abundant glossy curls.
-
-03:37.098 --> 03:39.820
-Jimmy, the same age as Ann, is stockier built.
-
-03:40.041 --> 03:41.602
-His bones and muscles are heavier.
-
-03:41.962 --> 03:45.265
-He walked sooner, partly because he had more chance to creep.
-
-03:47.867 --> 03:52.751
-All these preschool children have been well fed since birth and have good appetites.
-
-03:56.623 --> 03:59.224
-Agnes and Carl both two and a half years.
-
-03:59.764 --> 04:04.566
-Carl has a well-shaped head, glossy hair, and very good color.
-
-04:05.126 --> 04:10.568
-As he watches his dandelion sail away presently, see what fine teeth he has.
-
-04:11.068 --> 04:18.791
-How straight and even they are in his broad, well-developed jaw, which gives plenty of room for his second teeth.
-
-04:22.051 --> 04:27.192
-Joyous fairy like Patsy with Carly's beautifully developed child who rejoices in the activity.
-
-04:27.652 --> 04:29.913
-Patsy at four years knows her foods.
-
-04:30.313 --> 04:36.835
-She said to her aunt, I mustn't drink cocoa, tea, coffee, cola, nothing but milk.
-
-04:42.676 --> 04:46.237
-Patsy is going to serve milk to her little friends.
-
-04:47.317 --> 04:51.258
-Patsy and dark haired Marilyn are both four years old.
-
-05:02.408 --> 05:07.652
-Bobby Six and his sister Irene Four have well-grown bones and strong muscles.
-
-05:08.452 --> 05:10.153
-What good control they have of them.
-
-05:10.854 --> 05:15.737
-Bobby's perfectly developed chest and ribs give plenty of room for breathing.
-
-05:18.359 --> 05:24.744
-Bobby also, you will note, has well-spaced, even teeth in a well-developed jaw.
-
-05:26.980 --> 05:30.301
-Note these happy seven to 11-year-old children.
-
-05:31.061 --> 05:37.484
-Barbara, standing erect in the foreground calling to Joan, is well-developed, tall, with broad shoulders.
-
-05:38.344 --> 05:41.305
-Barbara and Joan are both seven years old.
-
-05:42.125 --> 05:45.967
-Joan, too, is well-developed, but short and stalky.
-
-05:46.847 --> 05:49.828
-This older group shows differences in size and build.
-
-05:50.128 --> 05:54.110
-Harriet, the little one in the center, is 11 years old, and Marilyn, the largest, 13.
-
-05:58.040 --> 06:01.062
-This is David, nine years old, and Don, twelve.
-
-06:01.442 --> 06:03.183
-Don is a thin, wiry type.
-
-06:03.723 --> 06:05.864
-David, whom we have met before, is heavy.
-
-06:12.248 --> 06:25.975
-Jean, eleven years old, shows beautiful development and excellent control of her muscles, and is a lovely-looking creature, because from infancy, Jean has had everything that helps to make a beautiful human being.
-
-06:32.601 --> 06:37.683
-Alan and his sister Betty illustrate the racehorse type with long, slender bones and muscles.
-
-06:38.243 --> 06:43.006
-Both are still growing rapidly and need plenty of the right kind of food and rest.
-
-06:44.146 --> 06:49.849
-Alan shows excellent control of his long, slender muscles, carries himself well and breathes well.
-
-06:50.509 --> 06:55.272
-Alan's charming smile as he chins shows vitality and joy in living.
-
-06:56.586 --> 07:01.171
-These children, helping mother pick vegetables, get many kinds of nourishment.
-
-07:01.731 --> 07:05.775
-Fresh air, sunshine, vegetables, eggs, too.
-
-07:07.136 --> 07:11.761
-Now, note the following necessary foods.
-
-07:14.003 --> 07:16.866
-Milk, the best bone building food for all ages.
-
-07:18.167 --> 07:20.610
-Oranges, an excellent source of vitamin C.
-
-07:26.008 --> 07:29.830
-Also cabbage and tomatoes, cabbage usually not for two years.
-
-07:30.911 --> 07:34.153
-Potato is a cheap source of iron, vitamin C, and energy.
-
-07:34.953 --> 07:37.995
-Green and yellow vegetables, strained of course for baby.
-
-07:43.559 --> 07:46.520
-Dried beans for blood and muscle, not for young children.
-
-07:48.382 --> 07:50.743
-This cut of beef costs little.
-
-07:51.373 --> 07:55.977
-Pig's liver at 12 cents a pound has all this cut has, plus all the vitamins.
-
-07:56.417 --> 07:57.758
-Strained liver for baby.
-
-07:58.639 --> 08:00.901
-Dried fruits strained, of course, for baby.
-
-08:04.945 --> 08:05.985
-Fresh fruits too.
-
-08:08.868 --> 08:10.790
-Bread whole-grained after two years.
-
-08:16.414 --> 08:19.497
-Whole-grained cereal like rolled oats with milk is fine for breakfast.
-
-08:20.341 --> 08:21.261
-Strain for baby.
-
-08:23.603 --> 08:25.884
-Eggs, one a day for each child.
-
-08:31.806 --> 08:35.168
-Salmon or other fat sea fish, a cheap form of protein food.
-
-08:35.648 --> 08:40.050
-The strained oil can take the place of some of your cod liver oil, even for baby.
-
-08:41.571 --> 08:46.853
-Bacon, an appetizing energy food, especially for older growing boys and girls.
-
-08:50.540 --> 08:52.181
-should give more than just energy.
-
-08:52.581 --> 08:55.363
-Molasses and brown sugar give some iron also.
-
-08:55.844 --> 08:56.964
-White sugar does not.
-
-09:00.046 --> 09:03.629
-Well-nurtured children become healthy and happy grown-ups.
-
-09:04.429 --> 09:20.180
-An abundant and properly balanced diet, based on the foods just seen, plus plenty of sunlight and play and a due portion of the milk of human kindness, makes children grow into healthy, well-balanced, efficient citizens.
-
-09:31.286 --> 09:35.067
-All the youngsters seen in these pictures have had such care.
-
-09:45.970 --> 09:50.992
-We would see to it that the boys and girls in our homes and on our playgrounds
-
-09:51.523 --> 10:06.536
-have the inexpensive yet priceless advantages these youngsters have had and are thus well fitted to take their places in the ranks of those who do the everyday work of the world, the work on which our national welfare depends.
-
-10:19.102 --> 10:20.624
-Oh, my God.
-
-10:52.411 --> 10:55.504
-This has been a presentation of the Library of Congress.
-
-10:56.106 --> 10:57.953
-Visit us at loc.gov.
-