Faster Whisperを使ってみる（GPUでの実行、Python、Node-RED）

py -3.10 -m venv pyenv310
cd pyenv310
.\Scripts\activate
pip install faster-whisper

[{"id":"35dcc8614a07234a","type":"inject","z":"22eb2b8f4786695c","name":"","props":[{"p":"payload"}],"repeat":"","crontab":"","once":false,"onceDelay":0.1,"topic":"","payload":"Hello, this is a sample voice. May I help you? I can talk to you by voice. My interface will be further extended.","payloadType":"str","x":1830,"y":2760,"wires":[["5d52af097769c75c"]]},{"id":"64b94f8f700be65e","type":"venv-exec","z":"22eb2b8f4786695c","name":"gTTS","venvconfig":"36c2cf6f351fdc6e","mode":"execute","executable":"gtts-cli.exe","arguments":"","x":1970,"y":2820,"wires":[["1fd3d1f5c03a5732"]]},{"id":"1fd3d1f5c03a5732","type":"debug","z":"22eb2b8f4786695c","name":"debug 427","active":true,"tosidebar":true,"console":false,"tostatus":false,"complete":"false","statusVal":"","statusType":"auto","x":2130,"y":2820,"wires":[]},{"id":"5d52af097769c75c","type":"template","z":"22eb2b8f4786695c","name":"","field":"payload","fieldType":"msg","format":"handlebars","syntax":"mustache","template":"'{{payload}}' --output voice.mp3","output":"str","x":1980,"y":2760,"wires":[["3dd3393f1f930dac","dee234e23e403bf1"]]},{"id":"3dd3393f1f930dac","type":"debug","z":"22eb2b8f4786695c","name":"debug 428","active":true,"tosidebar":true,"console":false,"tostatus":false,"complete":"false","statusVal":"","statusType":"auto","x":2150,"y":2720,"wires":[]},{"id":"dee234e23e403bf1","type":"change","z":"22eb2b8f4786695c","name":"","rules":[{"t":"change","p":"payload","pt":"msg","from":"\"","fromt":"str","to":"","tot":"str"}],"action":"","property":"","from":"","to":"","reg":false,"x":2180,"y":2760,"wires":[["64b94f8f700be65e"]]},{"id":"36c2cf6f351fdc6e","type":"venv-config","venvname":"AI","version":"3.8"}]

from faster_whisper import WhisperModel

model_size = "large-v3"
model = WhisperModel(model_size, device="cpu", compute_type="int8")

segments, info = model.transcribe("voice.mp3", beam_size=5)

print("Detected language '%s' with probability %f" % (info.language, info.language_probability))

for segment in segments:
    print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))

segments, info = model.transcribe("voice.mp3", beam_size=5, language="en")

pip uninstall torchvision
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126

from faster_whisper import WhisperModel

model_size = "large-v3"
model = WhisperModel(model_size, device="cuda", compute_type="int8")

segments, info = model.transcribe("voice.mp3", beam_size=5, language="en")

for segment in segments:
    print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))

[{"id":"d34b1c24f210765d","type":"comment","z":"22eb2b8f4786695c","name":"Faster Whisper","info":"","x":1800,"y":2720,"wires":[]},{"id":"00832978f0f1b880","type":"pip","z":"22eb2b8f4786695c","venvconfig":"6fee16ddf4b551d4","name":"","arg":"faster-whisper","action":"install","tail":false,"x":1930,"y":2780,"wires":[["cfb8692e8f555b75"]]},{"id":"bae2d94ba8963138","type":"inject","z":"22eb2b8f4786695c","name":"","props":[],"repeat":"","crontab":"","once":false,"onceDelay":0.1,"topic":"","x":1790,"y":2780,"wires":[["00832978f0f1b880"]]},{"id":"b8890ca52996e4b4","type":"debug","z":"22eb2b8f4786695c","name":"debug 427","active":true,"tosidebar":true,"console":false,"tostatus":false,"complete":"false","statusVal":"","statusType":"auto","x":2230,"y":2780,"wires":[]},{"id":"cfb8692e8f555b75","type":"pip","z":"22eb2b8f4786695c","venvconfig":"6fee16ddf4b551d4","name":"","arg":"torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126","action":"install","tail":false,"x":2070,"y":2780,"wires":[["b8890ca52996e4b4"]]},{"id":"2fdbcc97e5a92ab2","type":"inject","z":"22eb2b8f4786695c","name":"","props":[{"p":"model","v":"large-v3","vt":"str"},{"p":"voice_path","v":"voice.mp3","vt":"str"}],"repeat":"","crontab":"","once":false,"onceDelay":0.1,"topic":"","x":1790,"y":2840,"wires":[["8b29ca988c0d3af3"]]},{"id":"045ca9bc4dc2d245","type":"debug","z":"22eb2b8f4786695c","name":"debug 428","active":true,"tosidebar":true,"console":false,"tostatus":false,"complete":"false","statusVal":"","statusType":"auto","x":2090,"y":2840,"wires":[]},{"id":"8b29ca988c0d3af3","type":"venv","z":"22eb2b8f4786695c","venvconfig":"6fee16ddf4b551d4","name":"","code":"from faster_whisper import WhisperModel\n\nmodel_size = msg['model']\nmodel = WhisperModel(model_size, device=\"cuda\", compute_type=\"float16\")\n\nsegments, info = model.transcribe(msg['voice_path'], beam_size=5, language=\"en\")\n\nfor segment in segments:\n    print(\"[%.2fs -> %.2fs] %s\" % (segment.start, segment.end, segment.text))","continuous":false,"x":1930,"y":2840,"wires":[["045ca9bc4dc2d245"]]},{"id":"6fee16ddf4b551d4","type":"venv-config","venvname":"FasterWhisper","version":"3.10"}]