Mercurial

annotate dictation/main.py @ 282:6bd4a3990696 default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add cross-platform canvas orchestration and typed composer Enable native ARM dependencies, macOS Copilot orchestration, portable service supervision, CPU dictation, and a shared N-key speak-or-type composer. Co-authored-by: Copilot <[email protected]> Copilot-Session: f68442b1-fa8f-46a0-9689-81710613bbd4
author MrJuneJune <me@mrjunejune.com>
date Thu, 20 Aug 2026 20:45:25 -0700
parents 78699f810817
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
282
6bd4a3990696 Add cross-platform canvas orchestration and typed composer
MrJuneJune <me@mrjunejune.com>
parents: 275
diff changeset
40 compute_types = sorted(
6bd4a3990696 Add cross-platform canvas orchestration and typed composer
MrJuneJune <me@mrjunejune.com>
parents: 275
diff changeset
41 ctranslate2.get_supported_compute_types(config.device)
6bd4a3990696 Add cross-platform canvas orchestration and typed composer
MrJuneJune <me@mrjunejune.com>
parents: 275
diff changeset
42 )
275
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
43 if config.compute_type not in compute_types:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
44 raise RuntimeError(
282
6bd4a3990696 Add cross-platform canvas orchestration and typed composer
MrJuneJune <me@mrjunejune.com>
parents: 275
diff changeset
45 f"{config.compute_type} is unsupported on {config.device}; "
275
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
46 f"available: {', '.join(compute_types)}"
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
47 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
48 result = {
282
6bd4a3990696 Add cross-platform canvas orchestration and typed composer
MrJuneJune <me@mrjunejune.com>
parents: 275
diff changeset
49 "device": config.device,
275
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
50 "computeTypes": compute_types,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
51 "selectedComputeType": config.compute_type,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
52 "modelDir": str(config.model_dir),
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
53 "modelPresent": (config.model_dir / "model.bin").is_file(),
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
54 }
282
6bd4a3990696 Add cross-platform canvas orchestration and typed composer
MrJuneJune <me@mrjunejune.com>
parents: 275
diff changeset
55 if config.device == "cuda":
6bd4a3990696 Add cross-platform canvas orchestration and typed composer
MrJuneJune <me@mrjunejune.com>
parents: 275
diff changeset
56 result["cudaDeviceCount"] = ctranslate2.get_cuda_device_count()
6bd4a3990696 Add cross-platform canvas orchestration and typed composer
MrJuneJune <me@mrjunejune.com>
parents: 275
diff changeset
57 if config.device == "cuda" and result["cudaDeviceCount"] < 1:
275
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
58 raise RuntimeError("CTranslate2 did not detect a CUDA device")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
59 print(json.dumps(result, indent=2))
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
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
62 async def transcribe_file(config: DictationConfig, path: Path) -> None:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
63 from dictation.transcriber import FasterWhisperTranscriber
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 transcriber = FasterWhisperTranscriber(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
66 config.model_dir,
282
6bd4a3990696 Add cross-platform canvas orchestration and typed composer
MrJuneJune <me@mrjunejune.com>
parents: 275
diff changeset
67 config.device,
275
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
68 config.compute_type,
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 try:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
71 from av import open as av_open
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
72 from av.audio.resampler import AudioResampler
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
73 import numpy as np
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
74
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
75 samples = []
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
76 resampler = AudioResampler(format="s16", layout="mono", rate=16000)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
77 with av_open(str(path)) as container:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
78 for frame in container.decode(audio=0):
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
79 for converted in resampler.resample(frame):
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
80 samples.append(converted.to_ndarray().reshape(-1))
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
81 if not samples:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
82 raise RuntimeError("Audio file contains no decodable samples")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
83 audio = np.concatenate(samples).astype(np.float32) / 32768.0
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
84 result = await transcriber.transcribe(audio, final=True)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
85 print(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
86 json.dumps(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
87 {
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
88 "text": result.text,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
89 "language": result.language,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
90 "probability": result.probability,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
91 },
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
92 ensure_ascii=False,
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 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
95 finally:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
96 transcriber.close()
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
97
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
98
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
99 def main() -> None:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
100 parser = argparse.ArgumentParser()
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
101 parser.add_argument(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
102 "command",
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
103 choices=["server", "preflight", "download_model", "transcribe"],
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
104 )
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
105 parser.add_argument("path", nargs="?")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
106 args = parser.parse_args()
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
107 try:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
108 config = DictationConfig.from_environment()
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
109 if args.command == "download_model":
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
110 download_model(config)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
111 elif args.command == "preflight":
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
112 preflight(config)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
113 elif args.command == "transcribe":
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
114 if not args.path:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
115 parser.error("transcribe requires an audio file path")
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
116 asyncio.run(transcribe_file(config, Path(args.path)))
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
117 else:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
118 import uvicorn
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
119 from dictation.server import create_app
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
120
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
121 uvicorn.run(
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
122 create_app(config),
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
123 host=config.host,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
124 port=config.port,
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
125 access_log=False,
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 except Exception as error:
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
128 print(f"dictation: {error}", file=sys.stderr)
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
129 raise SystemExit(1) from error
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
130
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
131
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
132 if __name__ == "__main__":
78699f810817 Add Qwen3-VL and WebRTC dictation services
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
133 main()