Mercurial

annotate dictation/main.py @ 275:78699f810817

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add Qwen3-VL and WebRTC dictation services Add Bazel targets for the CUDA-backed Qwen3-VL server and a local WebRTC faster-whisper dictation service. Co-authored-by: Copilot <[email protected]> Copilot-Session: e3d8cb06-6c95-4ae0-9757-651d3796ab00
author MrJuneJune <me@mrjunejune.com>
date Mon, 17 Aug 2026 10:58:47 -0700
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
275
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
1 from __future__ import annotations
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
2
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
3 import argparse
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
4 import asyncio
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
5 import json
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
6 import os
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
7 from pathlib import Path
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
8 import sys
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
9
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
10 from dictation.config import (
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
11 DictationConfig,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
12 MODEL_REPOSITORY,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
13 MODEL_REVISION,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
14 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
15
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
16
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
17 def download_model(config: DictationConfig) -> None:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
18 from huggingface_hub import snapshot_download
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
19
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
20 config.model_dir.mkdir(parents=True, exist_ok=True)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
21 snapshot_download(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
22 repo_id=MODEL_REPOSITORY,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
23 revision=MODEL_REVISION,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
24 local_dir=config.model_dir,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
25 allow_patterns=[
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
26 "config.json",
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
27 "model.bin",
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
28 "tokenizer.json",
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
29 "vocabulary.txt",
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
30 ],
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
31 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
32 marker = config.model_dir / ".model-revision"
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
33 marker.write_text(f"{MODEL_REPOSITORY}@{MODEL_REVISION}\n", encoding="ascii")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
34 print(f"Model ready: {config.model_dir}")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
35
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
36
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
37 def preflight(config: DictationConfig) -> None:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
38 import ctranslate2
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
39
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
40 compute_types = sorted(ctranslate2.get_supported_compute_types("cuda"))
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
41 if config.compute_type not in compute_types:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
42 raise RuntimeError(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
43 f"{config.compute_type} is unsupported on CUDA; "
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
44 f"available: {', '.join(compute_types)}"
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
45 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
46 result = {
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
47 "cudaDeviceCount": ctranslate2.get_cuda_device_count(),
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
48 "computeTypes": compute_types,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
49 "selectedComputeType": config.compute_type,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
50 "modelDir": str(config.model_dir),
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
51 "modelPresent": (config.model_dir / "model.bin").is_file(),
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
52 }
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
53 if result["cudaDeviceCount"] < 1:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
54 raise RuntimeError("CTranslate2 did not detect a CUDA device")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
55 print(json.dumps(result, indent=2))
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
56
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
57
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
58 async def transcribe_file(config: DictationConfig, path: Path) -> None:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
59 from dictation.transcriber import FasterWhisperTranscriber
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
60
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
61 transcriber = FasterWhisperTranscriber(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
62 config.model_dir,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
63 config.compute_type,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
64 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
65 try:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
66 from av import open as av_open
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
67 from av.audio.resampler import AudioResampler
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
68 import numpy as np
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
69
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
70 samples = []
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
71 resampler = AudioResampler(format="s16", layout="mono", rate=16000)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
72 with av_open(str(path)) as container:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
73 for frame in container.decode(audio=0):
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
74 for converted in resampler.resample(frame):
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
75 samples.append(converted.to_ndarray().reshape(-1))
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
76 if not samples:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
77 raise RuntimeError("Audio file contains no decodable samples")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
78 audio = np.concatenate(samples).astype(np.float32) / 32768.0
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
79 result = await transcriber.transcribe(audio, final=True)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
80 print(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
81 json.dumps(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
82 {
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
83 "text": result.text,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
84 "language": result.language,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
85 "probability": result.probability,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
86 },
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
87 ensure_ascii=False,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
88 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
89 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
90 finally:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
91 transcriber.close()
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
92
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
93
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
94 def main() -> None:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
95 parser = argparse.ArgumentParser()
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
96 parser.add_argument(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
97 "command",
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
98 choices=["server", "preflight", "download_model", "transcribe"],
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
99 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
100 parser.add_argument("path", nargs="?")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
101 args = parser.parse_args()
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
102 try:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
103 config = DictationConfig.from_environment()
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
104 if args.command == "download_model":
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
105 download_model(config)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
106 elif args.command == "preflight":
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
107 preflight(config)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
108 elif args.command == "transcribe":
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
109 if not args.path:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
110 parser.error("transcribe requires an audio file path")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
111 asyncio.run(transcribe_file(config, Path(args.path)))
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
112 else:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
113 import uvicorn
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
114 from dictation.server import create_app
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
115
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
116 uvicorn.run(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
117 create_app(config),
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
118 host=config.host,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
119 port=config.port,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
120 access_log=False,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
121 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
122 except Exception as error:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
123 print(f"dictation: {error}", file=sys.stderr)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
124 raise SystemExit(1) from error
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
125
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
126
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
127 if __name__ == "__main__":
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
128 main()