every_holiday/VRCT
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[WIP/TEST] distil-wisperを削除/faster-wisperの処理を修正

Browse Source
This commit is contained in:
misyaguziya
2024-02-04 02:42:08 +09:00
parent 7aafce6e2e
commit 61a6eb792b
6 changed files with 21 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
models/transcription/transcription_transcriber.py
View File

@@ -5,7 +5,7 @@ from speech_recognition import Recognizer, AudioData, AudioFile
from datetime import timedelta
from pyaudiowpatch import get_sample_size, paInt16
from .transcription_languages import transcription_lang
from .transcription_whisper import getWhisperModel
from .transcription_whisper import getWhisperModel, checkWhisperWeight
import torch
import numpy as np
@@ -14,7 +14,7 @@ PHRASE_TIMEOUT = 3
MAX_PHRASES = 10
class AudioTranscriber:
def __init__(self, speaker, source, phrase_timeout, max_phrases, transcription_engine, whisper_weight_type=None, root=None):
def __init__(self, speaker, source, phrase_timeout, max_phrases, root=None, whisper_weight_type=None, ):
self.speaker = speaker
self.phrase_timeout = phrase_timeout
self.max_phrases = max_phrases
@@ -30,34 +30,37 @@ class AudioTranscriber:
"new_phrase": True,
"process_data_func": self.processSpeakerData if speaker else self.processSpeakerData
}
self.transcription_engine = transcription_engine
match self.transcription_engine:
case "Google":
self.audio_recognizer = Recognizer()
case "Whisper":
self.audio_recognizer = getWhisperModel(root, whisper_weight_type)
if whisper_weight_type is not None and root is not None and checkWhisperWeight(root, whisper_weight_type) is True:
self.whisper_model = getWhisperModel(root, whisper_weight_type)
else:
self.whisper_model = None
def transcribeAudioQueue(self, audio_queue, language, country):
def transcribeAudioQueue(self, audio_queue, language, country, transcription_engine):
audio, time_spoken = audio_queue.get()
self.updateLastSampleAndPhraseStatus(audio, time_spoken)
text = ''
try:
# Whisperが使用できない場合はGoogle Speech-to-Textを使用する
if transcription_engine == "Whisper":
if self.whisper_model is None:
transcription_engine = "Google"
audio_data = self.audio_sources["process_data_func"]()
match self.transcription_engine:
match transcription_engine:
case "Google":
text = self.audio_recognizer.recognize_google(audio_data, language=transcription_lang[language][country][self.transcription_engine])
text = self.audio_recognizer.recognize_google(audio_data, language=transcription_lang[language][country][transcription_engine])
case "Whisper":
audio_data = np.frombuffer(audio_data.get_raw_data(convert_rate=16000, convert_width=2), np.int16).flatten().astype(np.float32) / 32768.0
if isinstance(audio_data, torch.Tensor):
audio_data = audio_data.detach().numpy()
segments, _ = self.audio_recognizer.transcribe(
segments, _ = self.whisper_model.transcribe(
audio_data,
beam_size=5,
temperature=0.0,
log_prob_threshold=-0.8,
no_speech_threshold=0.6,
language=transcription_lang[language][country][self.transcription_engine],
language=transcription_lang[language][country][transcription_engine],
word_timestamps=False,
without_timestamps=True,
task="transcribe",

3
models/transcription/transcription_whisper.py
View File

@@ -15,9 +15,6 @@ _MODELS = {
"large-v1": "Systran/faster-whisper-large-v1",
"large-v2": "Systran/faster-whisper-large-v2",
"large-v3": "Systran/faster-whisper-large-v3",
"distil-small": "Systran/faster-distil-whisper-small.en",
"distil-medium": "Systran/faster-distil-whisper-medium.en",
"distil-large-v2": "Systran/faster-distil-whisper-large-v2"
}
_FILENAMES = [