Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
import os
import json
import torch
import logging
import concurrent.futures
import librosa
import torch.distributed as dist
from typing import Collection
import torch
try:
import torchaudio
except ImportError:
torchaudio = None
from torch import nn
import random
import re
import string
from funasr.tokenizer.cleaner import TextCleaner
from funasr.register import tables
@tables.register("preprocessor_classes", "TextPreprocessRemovePunctuation")
class TextPreprocessRemovePunctuation(nn.Module):
def __init__(self, **kwargs):
"""Initialize TextPreprocessRemovePunctuation.
Args:
**kwargs: Additional keyword arguments.
"""
super().__init__()
def forward(self, text, **kwargs):
# 定义英文标点符号
"""Forward pass for training.
Args:
text: Text tensor or string input.
**kwargs: Additional keyword arguments.
"""
en_punct = string.punctuation
# 定义中文标点符号(部分常用的)
cn_punct = "。?!,、;:“”‘’()《》【】…—~·"
# 合并英文和中文标点符号
all_punct = en_punct + cn_punct
# 创建正则表达式模式,匹配任何在all_punct中的字符
punct_pattern = re.compile("[{}]".format(re.escape(all_punct)))
# 使用正则表达式的sub方法替换掉这些字符
return punct_pattern.sub("", text)