deploy_model.py

'''
Author: wuxulong19950206 [email protected]
Date: 2024-03-11 19:46:42
LastEditors: wuxulong19950206 [email protected]
LastEditTime: 2024-03-11 21:02:36
FilePath: \apeech_rec\deploy_model.py
Description: 这是默认设置,请设置`customMade`, 打开koroFileHeader查看配置 进行设置: https://github.com/OBKoro1/koro1FileHeader/wiki/%E9%85%8D%E7%BD%AE
'''
import os

from speech_model import ModelSpeech
from model_zoo.speech_model.keras_backend import SpeechModel251BN
from speech_features import Spectrogram
from language_model3 import ModelLanguage

os.environ["CUDA_VISIBLE_DEVICES"] = "0"

AUDIO_LENGTH = 1600
AUDIO_FEATURE_LENGTH = 200
CHANNELS = 1
# 默认输出的拼音的表示大小是1428，即1427个拼音+1个空白块
OUTPUT_SIZE = 2884
sm251bn = SpeechModel251BN(
        input_shape=(AUDIO_LENGTH, AUDIO_FEATURE_LENGTH, CHANNELS),
        output_size=OUTPUT_SIZE
    )
def get_pretrained_model():

    
    feat = Spectrogram()
    ms = ModelSpeech(sm251bn, feat, max_label_length=64)
    if os.path.exists('./models/SpeechModel251bn_epoch12.model.h5'):
        print('*[提示] 加载预训练的声学模型')
    else:
        print('*[提示] 预训练的声学模型不存在')
    ms.load_model('./models/SpeechModel251bn_epoch12.model.h5')
    return ms
def decode(model,filename):

    res = model.recognize_speech_from_file(filename)
    print('*[提示] 声学模型语音识别结果：\n', res)
    return res
def not_use():
    ml = ModelLanguage('model_language')
    ml.load_model()
    str_pinyin = res
    res = ml.pinyin_to_text(str_pinyin)
    print('语音识别最终结果：\n', res)
chinese_models = {"chinese":'models\SpeechModel251bn_epoch12.model.h5'}
language_to_models = {
    "Chinese": list(chinese_models.keys())}