语音识别_Faster_Whisper_Base

faster_whisper base模型

frameworks:

• Pytorch license: Apache License 2.0 tasks:
• auto-speech-recognition

SDK下载

#安装ModelScope
pip install modelscope

#SDK模型下载
from modelscope import snapshot_download
model_dir = snapshot_download('yuanjian/kry_faster_whisper')

Git下载

#Git模型下载
git clone https://www.modelscope.cn/yuanjian/kry_faster_whisper.git

使用模型

#SDK模型下载
from modelscope import snapshot_download
model_dir = snapshot_download('yuanjian/kry_faster_whisper')

from faster_whisper import WhisperModel
import time

def transcribe(model, audio_path):
    st = time.time()
    segments, info = model.transcribe(audio_path)
    segments = list(segments)
    print('识别文件： ', audio_path, ' 耗时:', time.time() - st)
    for segment in segments:
        print("识别结果：[%s]" % (segment.text))

def test():
    st = time.time()
    print("load model ...")
    # Run on GPU with FP16
    model = WhisperModel(model_dir, device="cuda", compute_type="float16")
    # Run on GPU with INT8 
    # model = WhisperModel(model_dir, device="cuda", compute_type="int8_float16")
    # Run on CPU with INT8
    # model = WhisperModel(model_dir, device="cpu", compute_type="int8")
    print('load model cost:', time.time() - st)

    transcribe(model, "three_fangyan.m4a")



test()

如果您是本模型的贡献者，我们邀请您根据模型贡献文档，及时完善模型卡片内容。