mirror of
https://github.com/BoardWare-Genius/jarvis-models.git
synced 2025-12-13 16:53:24 +00:00
delete asr tts unnecessary files
This commit is contained in:
0Xiao0
@ -1,45 +0,0 @@
|
||||
from io import BytesIO
|
||||
from typing import Any, Coroutine
|
||||
|
||||
from fastapi import Request, Response, status
|
||||
from fastapi.responses import JSONResponse
|
||||
|
||||
from ..asr.rapid_paraformer.utils import read_yaml
|
||||
from ..asr.rapid_paraformer import RapidParaformer
|
||||
from .blackbox import Blackbox
|
||||
from injector import singleton, inject
|
||||
|
||||
@singleton
|
||||
class ASR(Blackbox):
|
||||
|
||||
@inject
|
||||
def __init__(self,path = ".env.yaml") -> None:
|
||||
config = read_yaml(path)
|
||||
self.paraformer = RapidParaformer(config)
|
||||
|
||||
def __call__(self, *args, **kwargs):
|
||||
return self.processing(*args, **kwargs)
|
||||
|
||||
async def processing(self, *args, **kwargs):
|
||||
data = args[0]
|
||||
results = self.paraformer([BytesIO(data)])
|
||||
if len(results) == 0:
|
||||
return None
|
||||
return results[0]
|
||||
|
||||
def valid(self, data: any) -> bool:
|
||||
if isinstance(data, bytes):
|
||||
return True
|
||||
return False
|
||||
|
||||
async def fast_api_handler(self, request: Request) -> Response:
|
||||
data = (await request.form()).get("audio")
|
||||
if data is None:
|
||||
# self.logger.warn("asr bag request","type", "fast_api_handler", "api", "asr")
|
||||
return JSONResponse(content={"error": "data is required"}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
d = await data.read()
|
||||
try:
|
||||
txt = await self.processing(d)
|
||||
except ValueError as e:
|
||||
return JSONResponse(content={"error": str(e)}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
return JSONResponse(content={"text": txt}, status_code=status.HTTP_200_OK)
|
||||
@ -1,103 +0,0 @@
|
||||
from io import BytesIO
|
||||
from typing import Any, Coroutine
|
||||
|
||||
from fastapi import Request, Response, status
|
||||
from fastapi.responses import JSONResponse
|
||||
|
||||
from funasr import AutoModel
|
||||
from funasr.utils.postprocess_utils import rich_transcription_postprocess
|
||||
from .blackbox import Blackbox
|
||||
from injector import singleton, inject
|
||||
|
||||
import tempfile
|
||||
|
||||
import os
|
||||
from ..configuration import SenseVoiceConf
|
||||
|
||||
from ..log.logging_time import logging_time
|
||||
import logging
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
@singleton
|
||||
class ASR(Blackbox):
|
||||
mode: str
|
||||
url: str
|
||||
speed: int
|
||||
device: str
|
||||
language: str
|
||||
speaker: str
|
||||
|
||||
@logging_time(logger=logger)
|
||||
def model_init(self, sensevoice_config: SenseVoiceConf) -> None:
|
||||
|
||||
model_dir = "/home/gpu/Workspace/Models/SenseVoice/SenseVoiceSmall"
|
||||
|
||||
self.speed = sensevoice_config.speed
|
||||
self.device = sensevoice_config.device
|
||||
self.language = sensevoice_config.language
|
||||
self.speaker = sensevoice_config.speaker
|
||||
self.device = sensevoice_config.device
|
||||
self.url = ''
|
||||
self.mode = sensevoice_config.mode
|
||||
self.asr = None
|
||||
self.speaker_ids = None
|
||||
os.environ['CUDA_VISIBLE_DEVICES'] = str(sensevoice_config.device)
|
||||
if self.mode == 'local':
|
||||
self.asr = AutoModel(
|
||||
model=model_dir,
|
||||
trust_remote_code=True,
|
||||
remote_code= "/home/gpu/Workspace/SenseVoice/model.py",
|
||||
vad_model="fsmn-vad",
|
||||
vad_kwargs={"max_single_segment_time": 30000},
|
||||
device="cuda:0",
|
||||
)
|
||||
|
||||
else:
|
||||
self.url = sensevoice_config.url
|
||||
logging.info('#### Initializing SenseVoiceASR Service in cuda:' + str(sensevoice_config.device) + ' mode...')
|
||||
|
||||
@inject
|
||||
def __init__(self, sensevoice_config: SenseVoiceConf) -> None:
|
||||
self.model_init(sensevoice_config)
|
||||
|
||||
def __call__(self, *args, **kwargs):
|
||||
return self.processing(*args, **kwargs)
|
||||
|
||||
async def processing(self, *args, **kwargs):
|
||||
data = args[0]
|
||||
# 创建一个临时文件
|
||||
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio_file:
|
||||
temp_audio_file.write(data)
|
||||
temp_audio_path = temp_audio_file.name
|
||||
res = self.asr.generate(
|
||||
input=temp_audio_path,
|
||||
cache={},
|
||||
language="auto", # "zh", "en", "yue", "ja", "ko", "nospeech"
|
||||
use_itn=True,
|
||||
batch_size_s=60,
|
||||
merge_vad=True, #
|
||||
merge_length_s=15,
|
||||
)
|
||||
# results = self.paraformer([BytesIO(data)])
|
||||
results = rich_transcription_postprocess(res[0]["text"])
|
||||
os.remove(temp_audio_path)
|
||||
if len(results) == 0:
|
||||
return None
|
||||
return results
|
||||
|
||||
def valid(self, data: any) -> bool:
|
||||
if isinstance(data, bytes):
|
||||
return True
|
||||
return False
|
||||
|
||||
async def fast_api_handler(self, request: Request) -> Response:
|
||||
data = (await request.form()).get("audio")
|
||||
if data is None:
|
||||
# self.logger.warn("asr bag request","type", "fast_api_handler", "api", "asr")
|
||||
return JSONResponse(content={"error": "data is required"}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
d = await data.read()
|
||||
try:
|
||||
txt = await self.processing(d)
|
||||
except ValueError as e:
|
||||
return JSONResponse(content={"error": str(e)}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
return JSONResponse(content={"text": txt}, status_code=status.HTTP_200_OK)
|
||||
@ -1,93 +0,0 @@
|
||||
import io
|
||||
import time
|
||||
|
||||
import requests
|
||||
from fastapi import Request, Response, status
|
||||
from fastapi.responses import JSONResponse
|
||||
from injector import inject
|
||||
from injector import singleton
|
||||
|
||||
from ..log.logging_time import logging_time
|
||||
|
||||
from ..configuration import CosyVoiceConf
|
||||
from .blackbox import Blackbox
|
||||
|
||||
import soundfile
|
||||
import pyloudnorm as pyln
|
||||
import sys
|
||||
sys.path.append('/home/gpu/Workspace/CosyVoice')
|
||||
from cosyvoice.cli.cosyvoice import CosyVoice
|
||||
from cosyvoice.utils.file_utils import load_wav
|
||||
import torchaudio
|
||||
|
||||
import os
|
||||
import logging
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
@singleton
|
||||
class CosyVoiceTTS(Blackbox):
|
||||
mode: str
|
||||
url: str
|
||||
speed: int
|
||||
device: str
|
||||
language: str
|
||||
speaker: str
|
||||
|
||||
@logging_time(logger=logger)
|
||||
def model_init(self, cosyvoice_config: CosyVoiceConf) -> None:
|
||||
self.speed = cosyvoice_config.speed
|
||||
self.device = cosyvoice_config.device
|
||||
self.language = cosyvoice_config.language
|
||||
self.speaker = cosyvoice_config.speaker
|
||||
self.device = cosyvoice_config.device
|
||||
self.url = ''
|
||||
self.mode = cosyvoice_config.mode
|
||||
self.cosyvoicetts = None
|
||||
self.speaker_ids = None
|
||||
os.environ['CUDA_VISIBLE_DEVICES'] = str(cosyvoice_config.device)
|
||||
if self.mode == 'local':
|
||||
self.cosyvoicetts = CosyVoice('/home/gpu/Workspace/Models/CosyVoice/pretrained_models/CosyVoice-300M')
|
||||
|
||||
else:
|
||||
self.url = cosyvoice_config.url
|
||||
logging.info('#### Initializing CosyVoiceTTS Service in cuda:' + str(cosyvoice_config.device) + ' mode...')
|
||||
|
||||
@inject
|
||||
def __init__(self, cosyvoice_config: CosyVoiceConf) -> None:
|
||||
self.model_init(cosyvoice_config)
|
||||
|
||||
def __call__(self, *args, **kwargs):
|
||||
return self.processing(*args, **kwargs)
|
||||
|
||||
def valid(self, *args, **kwargs) -> bool:
|
||||
text = args[0]
|
||||
return isinstance(text, str)
|
||||
|
||||
@logging_time(logger=logger)
|
||||
def processing(self, *args, **kwargs) -> io.BytesIO | bytes:
|
||||
text = args[0]
|
||||
current_time = time.time()
|
||||
if self.mode == 'local':
|
||||
audio = self.cosyvoicetts.inference_sft(text, self.language)
|
||||
f = io.BytesIO()
|
||||
soundfile.write(f, audio['tts_speech'].cpu().numpy().squeeze(0), 22050, format='wav')
|
||||
f.seek(0)
|
||||
print("#### CosyVoiceTTS Service consume - local : ", (time.time() - current_time))
|
||||
return f.read()
|
||||
else:
|
||||
message = {
|
||||
"text": text
|
||||
}
|
||||
response = requests.post(self.url, json=message)
|
||||
print("#### CosyVoiceTTS Service consume - docker : ", (time.time()-current_time))
|
||||
return response.content
|
||||
|
||||
async def fast_api_handler(self, request: Request) -> Response:
|
||||
try:
|
||||
data = await request.json()
|
||||
except:
|
||||
return JSONResponse(content={"error": "json parse error"}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
text = data.get("text")
|
||||
if text is None:
|
||||
return JSONResponse(content={"error": "text is required"}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
return Response(content=self.processing(text), media_type="audio/wav", headers={"Content-Disposition": "attachment; filename=audio.wav"})
|
||||
@ -1,108 +0,0 @@
|
||||
import io
|
||||
import time
|
||||
|
||||
import requests
|
||||
from fastapi import Request, Response, status
|
||||
from fastapi.responses import JSONResponse
|
||||
from injector import inject
|
||||
from injector import singleton
|
||||
|
||||
from ..log.logging_time import logging_time
|
||||
|
||||
from ..configuration import MeloConf
|
||||
from .blackbox import Blackbox
|
||||
|
||||
import soundfile
|
||||
import pyloudnorm as pyln
|
||||
from melo.api import TTS
|
||||
|
||||
import logging
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
@singleton
|
||||
class MeloTTS(Blackbox):
|
||||
mode: str
|
||||
url: str
|
||||
speed: int
|
||||
device: str
|
||||
language: str
|
||||
speaker: str
|
||||
|
||||
@logging_time(logger=logger)
|
||||
def model_init(self, melo_config: MeloConf) -> None:
|
||||
self.speed = melo_config.speed
|
||||
self.device = melo_config.device
|
||||
self.language = melo_config.language
|
||||
self.speaker = melo_config.speaker
|
||||
self.device = melo_config.device
|
||||
self.url = ''
|
||||
self.mode = melo_config.mode
|
||||
self.melotts = None
|
||||
self.speaker_ids = None
|
||||
if self.mode == 'local':
|
||||
self.melotts = TTS(language=self.language, device=self.device)
|
||||
self.speaker_ids = self.melotts.hps.data.spk2id
|
||||
else:
|
||||
self.url = melo_config.url
|
||||
logging.info('#### Initializing MeloTTS Service in ' + self.device + ' mode...')
|
||||
|
||||
@inject
|
||||
def __init__(self, melo_config: MeloConf) -> None:
|
||||
self.model_init(melo_config)
|
||||
|
||||
def __call__(self, *args, **kwargs):
|
||||
return self.processing(*args, **kwargs)
|
||||
|
||||
def valid(self, *args, **kwargs) -> bool:
|
||||
text = args[0]
|
||||
return isinstance(text, str)
|
||||
|
||||
@logging_time(logger=logger)
|
||||
def processing(self, *args, **kwargs) -> io.BytesIO | bytes:
|
||||
text = args[0]
|
||||
current_time = time.time()
|
||||
if self.mode == 'local':
|
||||
audio = self.melotts.tts_to_file(text, self.speaker_ids[self.speaker], speed=self.speed)
|
||||
f = io.BytesIO()
|
||||
soundfile.write(f, audio, 44100, format='wav')
|
||||
f.seek(0)
|
||||
# print("#### MeloTTS Service consume - local : ", (time.time() - current_time))
|
||||
# return f.read()
|
||||
|
||||
|
||||
# Read the audio data from the buffer
|
||||
data, rate = soundfile.read(f, dtype='float32')
|
||||
|
||||
# Peak normalization
|
||||
peak_normalized_audio = pyln.normalize.peak(data, -1.0)
|
||||
|
||||
# Integrated loudness normalization
|
||||
meter = pyln.Meter(rate)
|
||||
loudness = meter.integrated_loudness(peak_normalized_audio)
|
||||
loudness_normalized_audio = pyln.normalize.loudness(peak_normalized_audio, loudness, -12.0)
|
||||
|
||||
# Write the loudness normalized audio to an in-memory buffer
|
||||
normalized_audio_buffer = io.BytesIO()
|
||||
soundfile.write(normalized_audio_buffer, loudness_normalized_audio, rate, format='wav')
|
||||
normalized_audio_buffer.seek(0)
|
||||
|
||||
print("#### MeloTTS Service consume - local : ", (time.time() - current_time))
|
||||
return normalized_audio_buffer.read()
|
||||
|
||||
else:
|
||||
message = {
|
||||
"text": text
|
||||
}
|
||||
response = requests.post(self.url, json=message)
|
||||
print("#### MeloTTS Service consume - docker : ", (time.time()-current_time))
|
||||
return response.content
|
||||
|
||||
async def fast_api_handler(self, request: Request) -> Response:
|
||||
try:
|
||||
data = await request.json()
|
||||
except:
|
||||
return JSONResponse(content={"error": "json parse error"}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
text = data.get("text")
|
||||
if text is None:
|
||||
return JSONResponse(content={"error": "text is required"}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
return Response(content=self.processing(text), media_type="audio/wav", headers={"Content-Disposition": "attachment; filename=audio.wav"})
|
||||
@ -1,40 +0,0 @@
|
||||
import io
|
||||
import time
|
||||
from ntpath import join
|
||||
|
||||
from fastapi import Request, Response, status
|
||||
from fastapi.responses import JSONResponse
|
||||
from .blackbox import Blackbox
|
||||
from ..tts.tts_service import TTService
|
||||
from injector import singleton
|
||||
|
||||
@singleton
|
||||
class TTS(Blackbox):
|
||||
|
||||
def __init__(self, *args, **kwargs) -> None:
|
||||
self.tts_service = TTService("yunfeineo")
|
||||
|
||||
def __call__(self, *args, **kwargs):
|
||||
return self.processing(*args, **kwargs)
|
||||
|
||||
def processing(self, *args, **kwargs) -> io.BytesIO:
|
||||
text = args[0]
|
||||
current_time = time.time()
|
||||
audio = self.tts_service.read(text)
|
||||
print("#### TTS Service consume : ", (time.time()-current_time))
|
||||
return audio
|
||||
|
||||
def valid(self, *args, **kwargs) -> bool:
|
||||
text = args[0]
|
||||
return isinstance(text, str)
|
||||
|
||||
async def fast_api_handler(self, request: Request) -> Response:
|
||||
try:
|
||||
data = await request.json()
|
||||
except:
|
||||
return JSONResponse(content={"error": "json parse error"}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
text = data.get("text")
|
||||
if text is None:
|
||||
return JSONResponse(content={"error": "text is required"}, status_code=status.HTTP_400_BAD_REQUEST)
|
||||
by = self.processing(text)
|
||||
return Response(content=by.read(), media_type="audio/wav", headers={"Content-Disposition": "attachment; filename=audio.wav"})
|
||||
Reference in New Issue
Block a user