josh/ytplayer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add presenter view, stats, lyrics worker, watch party with voice chat, timestamp sharing, soundbites, notes, transcript search, PiP, EQ, sleep fade, gestures and external players; fix the crossfade end-of-song race

Browse Source
This commit is contained in:
Jonathan Sykes
2026-09-20 00:07:12 +08:00
parent ca07e1fdf1
commit 9c1ec4ac51
17 changed files with 2912 additions and 111 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
scripts/lyrics/.dockerignore Normal file
View File

@@ -0,0 +1 @@
__pycache__/

15
scripts/lyrics/Dockerfile Normal file
View File

@@ -0,0 +1,15 @@
# Lyrics worker — transcribes saved songs that have no lyrics yet, one at a
# time, with faster-whisper on CPU (no API keys). Talks to the ytplayer
# service over the compose network only; see scripts/lyrics/auto_lyrics.py.
FROM python:3.12-slim
ENV PYTHONUNBUFFERED=1 \
HF_HOME=/models \
PIP_NO_CACHE_DIR=1
WORKDIR /app
COPY requirements.txt .
RUN pip install -r requirements.txt
COPY auto_lyrics.py .
# Model weights (downloaded on first use) and the worker's memory of
# instrumentals/failures live on volumes, so rebuilds don't re-download.
VOLUME ["/models", "/data"]
CMD ["sh", "-c", "exec python auto_lyrics.py --missing --watch ${WATCH_SECONDS:-300} --state /data/state.json --model ${WHISPER_MODEL:-large-v3-turbo} --threads ${WHISPER_THREADS:-2}"]

147
scripts/lyrics/auto_lyrics.py
View File

@@ -121,9 +121,83 @@ def main():
ap.add_argument('--threads', type=int, default=os.cpu_count() or 4)
ap.add_argument('--dry-run', action='store_true', help='transcribe and print, do not upload')
ap.add_argument('--min-words', type=int, default=25, help='fewer words = treat as instrumental')
ap.add_argument('--watch', type=int, default=0, metavar='SECONDS',
help='keep running: re-check for songs without lyrics every SECONDS (worker mode)')
ap.add_argument('--state', default='', help='JSON file remembering instrumentals/failures (worker mode)')
args = ap.parse_args()
if args.watch:
return watch(args)
api = Api(args.base, os.environ.get('YTP_TOKEN'), os.environ.get('YTP_ADMIN_PASSWORD'))
run_once(args, Api(args.base, os.environ.get('YTP_TOKEN'), os.environ.get('YTP_ADMIN_PASSWORD')))
def load_state(path):
try:
with open(path) as f:
return json.load(f)
except (OSError, ValueError):
return {}
def save_state(path, state):
if not path:
return
tmp = path + '.tmp'
with open(tmp, 'w') as f:
json.dump(state, f)
os.replace(tmp, path)
def watch(args):
"""Worker mode: poll for saved songs without lyrics and transcribe them one
at a time, forever. Runs at low CPU priority; the state file remembers
instrumentals (never retried) and failures (retried with backoff)."""
try:
os.nice(10)
except OSError:
pass
token, password = os.environ.get('YTP_TOKEN'), os.environ.get('YTP_ADMIN_PASSWORD')
if not (token or password):
print('lyrics worker: no YTP_TOKEN set (LYRICS_WORKER_TOKEN in the server env) — idle', flush=True)
while True:
time.sleep(3600)
args.missing, args.ids, args.overwrite, args.dry_run = True, None, False, False
model = None
while True:
try:
api = Api(args.base, token, password)
st, r = api.call('GET', '/api/admin/media')
if st != 200:
raise RuntimeError(f'listing failed ({st}): {r.get("error")}')
state = load_state(args.state)
now = time.time()
todo = [m for m in r['media'] if not m['lyricsLines']
and state.get(m['id'], {}).get('status') != 'instrumental'
and state.get(m['id'], {}).get('retry_at', 0) <= now]
if todo:
if model is None:
from faster_whisper import WhisperModel
print(f'lyrics worker: loading {args.model}', flush=True)
model = WhisperModel(args.model, device='cpu', compute_type='int8', cpu_threads=args.threads)
m = todo[0] # one song per cycle keeps the worker's footprint small
result = transcribe_one(args, api, model, m['id'])
entry = state.get(m['id'], {})
if result.startswith('instrumental'):
state[m['id']] = {'status': 'instrumental', 'at': now}
elif result.startswith('saved') or result.startswith('skip'):
state.pop(m['id'], None)
else:
fails = entry.get('fails', 0) + 1
state[m['id']] = {'status': 'failed', 'fails': fails, 'retry_at': now + min(86400, 900 * 2 ** fails), 'error': result[:200]}
save_state(args.state, state)
print(f'lyrics worker: {m["id"]} ({m.get("title", "")[:60]}): {result}', flush=True)
continue # straight on to the next song
except Exception as e: # never die: the next cycle retries
print(f'lyrics worker: {e}', flush=True)
time.sleep(args.watch)
def run_once(args, api):
if args.missing:
st, r = api.call('GET', '/api/admin/media')
if st != 200:
@@ -137,44 +211,43 @@ def main():
from faster_whisper import WhisperModel # imported late: listing works without it
model = WhisperModel(args.model, device='cpu', compute_type='int8', cpu_threads=args.threads)
summary = []
for vid in todo:
st, cur = api.call('GET', f'/api/notes/{vid}')
live = (cur or {}).get('lyrics') if st == 200 else None
if live and live['data']['lines'] and not args.overwrite:
summary.append((vid, 'skip: has lyrics'))
continue
st, audio = api.call('GET', f'/api/media/{vid}?a=1', raw=True)
if st != 200:
summary.append((vid, f'no cached audio ({st})'))
continue
with tempfile.NamedTemporaryFile(suffix='.m4a') as f:
f.write(audio)
f.flush()
t0 = time.time()
# vad_filter must stay OFF: it classifies sung music as non-speech
# and silently drops the whole song.
segs, info = model.transcribe(f.name, language=args.language, word_timestamps=True, vad_filter=False,
beam_size=5, condition_on_previous_text=False)
words = [{'text': w.word.strip(), 'start': w.start, 'end': w.end}
for s in segs for w in (s.words or []) if w.word.strip()]
took = time.time() - t0
if len(words) < args.min_words:
summary.append((vid, f'instrumental? only {len(words)} words — skipped'))
continue
lines = segment(words)
doc = {'lines': lines, 'tags': ['auto-transcribed (whisper)'], 'offset': 0}
head = ' / '.join(l['text'] for l in lines[:3])
print(f'{vid}: {len(lines)} lines, lang={info.language}, {took:.0f}s for {info.duration:.0f}s audio | {head[:100]}', flush=True)
if args.dry_run:
print(json.dumps(doc, ensure_ascii=False)[:2000])
summary.append((vid, f'dry-run {len(lines)} lines'))
continue
body = {'data': doc, 'baseRev': live['rev'] if live else 0}
st, r = api.call('PUT', f'/api/notes/{vid}/lyrics', body)
summary.append((vid, f'saved rev {r.get("rev")}' if st == 200 else f'upload failed {st}: {r.get("error")}'))
summary = [(vid, transcribe_one(args, api, model, vid)) for vid in todo]
print('\n'.join(f'{v} {s}' for v, s in summary))
def transcribe_one(args, api, model, vid):
"""Transcribe one saved song and upload it. Returns a one-line result."""
st, cur = api.call('GET', f'/api/notes/{vid}')
live = (cur or {}).get('lyrics') if st == 200 else None
if live and live['data']['lines'] and not args.overwrite:
return 'skip: has lyrics'
st, audio = api.call('GET', f'/api/media/{vid}?a=1', raw=True)
if st != 200:
return f'no cached audio ({st})'
with tempfile.NamedTemporaryFile(suffix='.m4a') as f:
f.write(audio)
f.flush()
t0 = time.time()
# vad_filter must stay OFF: it classifies sung music as non-speech
# and silently drops the whole song.
segs, info = model.transcribe(f.name, language=args.language, word_timestamps=True, vad_filter=False,
beam_size=5, condition_on_previous_text=False)
words = [{'text': w.word.strip(), 'start': w.start, 'end': w.end}
for s in segs for w in (s.words or []) if w.word.strip()]
took = time.time() - t0
if len(words) < args.min_words:
return f'instrumental? only {len(words)} words — skipped'
lines = segment(words)
doc = {'lines': lines, 'tags': ['auto-transcribed (whisper)'], 'offset': 0}
head = ' / '.join(l['text'] for l in lines[:3])
print(f'{vid}: {len(lines)} lines, lang={info.language}, {took:.0f}s for {info.duration:.0f}s audio | {head[:100]}', flush=True)
if args.dry_run:
print(json.dumps(doc, ensure_ascii=False)[:2000])
return f'dry-run {len(lines)} lines'
body = {'data': doc, 'baseRev': live['rev'] if live else 0}
st, r = api.call('PUT', f'/api/notes/{vid}/lyrics', body)
return f'saved rev {r.get("rev")}' if st == 200 else f'upload failed {st}: {r.get("error")}'
if __name__ == '__main__':
main()