ffmpeg提取视频中的音频-pcm
ffmpeg -y -i input.mp4 -vn -codec copy out.m4a
ffmpeg -i out.m4a -f segment -segment_time 55 -c copy out.m4a%03d.m4a
# 提取视频中的音频 mp4 - m4a
for %i in (*.mp4) do ffmpeg -i %i -vn -codec copy %i-out.m4a
# 音频按时间分割 - m4a
for %i in (*.m4a) do ffmpeg -i %i -f segment -segment_time 55 -c copy %i-%03d-.m4a
# 批量转换音频格式为pcm(参数设置)
for %i in (*-.m4a) do ffmpeg -i %i -acodec pcm_s16le -f s16le -ac 1 -ar 16000 %i-new.pcm
python+aip音频转文字
import os
from aip import AipSpeech
from mydocx import *
from tqdm import tqdm
class AudioToText:
''' 利用 FFmpeg 提取视频中的语音并转换为文本文档'''
def __init__(self):
self.document = Document()
self.document.styles['Normal'].font.name = u'微软雅黑'
self.document.styles['Normal']._element.rPr.rFonts.set(qn('w:eastAsia'), u'微软雅黑')
""" 你的 APPID AK SK """
APP_ID = '*'
API_KEY = '*'
SECRET_KEY = '*'
self.client = AipSpeech(APP_ID, API_KEY, SECRET_KEY)
def get_path_files(self, path, type='.pcm'):
'''
获取目录下所有的 pcm文件
:param path: 目录地址
:param type: 文件类型
:return: 包含文件路径的列表
'''
files = os.listdir(path)
new = []
for i in files:
name, ty = os.path.splitext(i)
if ty == type:
pathFile = os.path.join(path,i)
new.append(pathFile)
return new
def get_file_content(self, filePath):
'''
文件读取 content
:param filePath: 文件包含路径
:return:content
'''
with open(filePath, 'rb') as fp:
return fp.read()
def read_pcm(self, filename):
'''
识别pcm语音文件 转文字
:param filename: 文件名称
:return: 识别结果
'''
result = self.client.asr(self.get_file_content(filename), 'pcm', 16000, {'dev_pid': 1537,})
# baidu-api没想到这么脆弱！
try:
if result['err_msg'] == 'success.':
word = result['result'][0] + "\n\n"
else:
word = str(result['err_no']) + str(result['err_msg']) + "\n\n"
except:
word = str(result) + "\n\n"
# print(word)
return word,result['err_no']
def GoWalkPath(self, path):
'''
获取所有文件夹的地址
'''
videoPathList = []
for f_path, dir_name, f_names in os.walk(path):
if f_path != path:
videoPathList.append(f_path)
# print(f_path)
return videoPathList
def main(self, path):
'''
运行语音转文字并保持word文档
'''
pathList = self.GoWalkPath(path)
if len(pathList) > 0:
for _path in pathList:
files = self.get_path_files(_path)
for i in tqdm(range(0, len(files))):
wd,err = self.read_pcm(files[i])
if err ==0:
totle_level(self.document, files[i])
body(self.document, wd )
else:
body(self.document, wd, True)
try:
name = _path.split('\\')[-1]
_filePath = os.path.join(_path, name)
saveFile(self.document, _filePath)
except Exception as e:
print(e)
else:
print("目录下无文件夹")
if __name__ == '__main__':
# pcm 音频路径
path = "video"
t = AudioToText()
t.main(path)
