Spaces:
Sleeping
Sleeping
‘dalvqw’
commited on
Commit
•
dcd5f79
1
Parent(s):
303e4dd
增加批量总结音视频的功能
Browse files- crazy_functional.py +10 -0
- crazy_functions/crazy_utils.py +37 -0
- crazy_functions/总结音视频.py +138 -0
crazy_functional.py
CHANGED
@@ -246,5 +246,15 @@ def get_crazy_functions():
|
|
246 |
"Function": HotReload(图片生成)
|
247 |
},
|
248 |
})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
249 |
###################### 第n组插件 ###########################
|
250 |
return function_plugins
|
|
|
246 |
"Function": HotReload(图片生成)
|
247 |
},
|
248 |
})
|
249 |
+
from crazy_functions.总结音视频 import 总结音视频
|
250 |
+
function_plugins.update({
|
251 |
+
"批量总结音视频(输入路径或上传压缩包)": {
|
252 |
+
"Color": "stop",
|
253 |
+
"AsButton": False,
|
254 |
+
"AdvancedArgs": True,
|
255 |
+
"ArgsReminder": "调用openai api 使用whisper-1模型, 目前支持的格式:mp4, m4a, wav, mpga, mpeg, mp3, 此处无需输入参数",
|
256 |
+
"Function": HotReload(总结音视频)
|
257 |
+
}
|
258 |
+
})
|
259 |
###################### 第n组插件 ###########################
|
260 |
return function_plugins
|
crazy_functions/crazy_utils.py
CHANGED
@@ -606,3 +606,40 @@ def get_files_from_everything(txt, type): # type='.md'
|
|
606 |
success = False
|
607 |
|
608 |
return success, file_manifest, project_folder
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
606 |
success = False
|
607 |
|
608 |
return success, file_manifest, project_folder
|
609 |
+
|
610 |
+
|
611 |
+
def split_audio_file(filename, split_duration=1000):
|
612 |
+
"""
|
613 |
+
根据给定的切割时长将音频文件切割成多个片段。
|
614 |
+
|
615 |
+
Args:
|
616 |
+
filename (str): 需要被切割的音频文件名。
|
617 |
+
split_duration (int, optional): 每个切割音频片段的时长(以秒为单位)。默认值为1000。
|
618 |
+
|
619 |
+
Returns:
|
620 |
+
filelist (list): 一个包含所有切割音频片段文件路径的列表。
|
621 |
+
|
622 |
+
"""
|
623 |
+
from moviepy.editor import AudioFileClip
|
624 |
+
import os
|
625 |
+
os.makedirs('gpt_log/mp3/cut/', exist_ok=True) # 创建存储切割音频的文件夹
|
626 |
+
|
627 |
+
# 读取音频文件
|
628 |
+
audio = AudioFileClip(filename)
|
629 |
+
|
630 |
+
# 计算文件总时长和切割点
|
631 |
+
total_duration = audio.duration
|
632 |
+
split_points = list(range(0, int(total_duration), split_duration))
|
633 |
+
split_points.append(int(total_duration))
|
634 |
+
filelist = []
|
635 |
+
|
636 |
+
# 切割音频文件
|
637 |
+
for i in range(len(split_points) - 1):
|
638 |
+
start_time = split_points[i]
|
639 |
+
end_time = split_points[i + 1]
|
640 |
+
split_audio = audio.subclip(start_time, end_time)
|
641 |
+
split_audio.write_audiofile(f"gpt_log/mp3/cut/{filename[0]}_{i}.mp3")
|
642 |
+
filelist.append(f"gpt_log/mp3/cut/{filename[0]}_{i}.mp3")
|
643 |
+
|
644 |
+
audio.close()
|
645 |
+
return filelist
|
crazy_functions/总结音视频.py
ADDED
@@ -0,0 +1,138 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from toolbox import CatchException, report_execption, select_api_key, update_ui, write_results_to_file
|
2 |
+
from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive, split_audio_file
|
3 |
+
|
4 |
+
|
5 |
+
def AnalyAudio(file_manifest, llm_kwargs, chatbot, history):
|
6 |
+
import os, requests
|
7 |
+
from moviepy.editor import AudioFileClip
|
8 |
+
from request_llm.bridge_all import model_info
|
9 |
+
|
10 |
+
# 设置OpenAI密钥和模型
|
11 |
+
api_key = select_api_key(llm_kwargs['api_key'], llm_kwargs['llm_model'])
|
12 |
+
chat_endpoint = model_info[llm_kwargs['llm_model']]['endpoint']
|
13 |
+
|
14 |
+
whisper_endpoint = chat_endpoint.replace('chat/completions', 'audio/transcriptions')
|
15 |
+
url = whisper_endpoint
|
16 |
+
headers = {
|
17 |
+
'Authorization': f"Bearer {api_key}"
|
18 |
+
}
|
19 |
+
|
20 |
+
os.makedirs('gpt_log/mp3/', exist_ok=True)
|
21 |
+
for index, fp in enumerate(file_manifest):
|
22 |
+
audio_history = []
|
23 |
+
# 提取文件扩展名
|
24 |
+
ext = os.path.splitext(fp)[1]
|
25 |
+
# 提取视频中的音频
|
26 |
+
if ext not in [".mp3", ".wav", ".m4a", ".mpga"]:
|
27 |
+
audio_clip = AudioFileClip(fp)
|
28 |
+
audio_clip.write_audiofile(f'gpt_log/mp3/output{index}.mp3')
|
29 |
+
fp = f'gpt_log/mp3/output{index}.mp3'
|
30 |
+
# 调用whisper模型音频转文字
|
31 |
+
voice = split_audio_file(fp)
|
32 |
+
for j, i in enumerate(voice):
|
33 |
+
with open(i, 'rb') as f:
|
34 |
+
file_content = f.read() # 读取文件内容到内存
|
35 |
+
files = {
|
36 |
+
'file': (os.path.basename(i), file_content),
|
37 |
+
}
|
38 |
+
data = {
|
39 |
+
"model": "whisper-1",
|
40 |
+
'response_format': "text"
|
41 |
+
}
|
42 |
+
response = requests.post(url, headers=headers, files=files, data=data).text
|
43 |
+
|
44 |
+
i_say = f'请对下面的文章片段做概述,文章内容是 ```{response}```'
|
45 |
+
i_say_show_user = f'第{index + 1}段音频的第{j + 1} / {len(voice)}片段。'
|
46 |
+
gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
|
47 |
+
inputs=i_say,
|
48 |
+
inputs_show_user=i_say_show_user,
|
49 |
+
llm_kwargs=llm_kwargs,
|
50 |
+
chatbot=chatbot,
|
51 |
+
history=[],
|
52 |
+
sys_prompt="总结文章。"
|
53 |
+
)
|
54 |
+
|
55 |
+
chatbot[-1] = (i_say_show_user, gpt_say)
|
56 |
+
history.extend([i_say_show_user, gpt_say])
|
57 |
+
audio_history.extend([i_say_show_user, gpt_say])
|
58 |
+
|
59 |
+
# 已经对该文章的所有片段总结完毕,如果文章被切分了,
|
60 |
+
result = "".join(audio_history)
|
61 |
+
if len(audio_history) > 1:
|
62 |
+
i_say = f"根据以上的对话,使用中文总结文章{result}的主要内容。"
|
63 |
+
i_say_show_user = f'第{index + 1}段音频的主要内容:'
|
64 |
+
gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
|
65 |
+
inputs=i_say,
|
66 |
+
inputs_show_user=i_say_show_user,
|
67 |
+
llm_kwargs=llm_kwargs,
|
68 |
+
chatbot=chatbot,
|
69 |
+
history=audio_history,
|
70 |
+
sys_prompt="总结文章。"
|
71 |
+
)
|
72 |
+
|
73 |
+
history.extend([i_say, gpt_say])
|
74 |
+
audio_history.extend([i_say, gpt_say])
|
75 |
+
|
76 |
+
res = write_results_to_file(history)
|
77 |
+
chatbot.append((f"第{index + 1}段音频完成了吗?", res))
|
78 |
+
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
79 |
+
|
80 |
+
# 删除中间文件夹
|
81 |
+
import shutil
|
82 |
+
shutil.rmtree('gpt_log/mp3')
|
83 |
+
res = write_results_to_file(history)
|
84 |
+
chatbot.append(("所有音频都总结完成了吗?", res))
|
85 |
+
yield from update_ui(chatbot=chatbot, history=history)
|
86 |
+
|
87 |
+
|
88 |
+
@CatchException
|
89 |
+
def 总结音视频(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, WEB_PORT):
|
90 |
+
import glob, os
|
91 |
+
|
92 |
+
# 基本信息:功能、贡献者
|
93 |
+
chatbot.append([
|
94 |
+
"函数插件功能?",
|
95 |
+
"总结音视频内容,函数插件贡献者: dalvqw"])
|
96 |
+
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
97 |
+
|
98 |
+
try:
|
99 |
+
from moviepy.editor import AudioFileClip
|
100 |
+
except:
|
101 |
+
report_execption(chatbot, history,
|
102 |
+
a=f"解析项目: {txt}",
|
103 |
+
b=f"导入软件依赖失败。使用该模块需要额外依赖,安装方法```pip install --upgrade moviepy```。")
|
104 |
+
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
105 |
+
return
|
106 |
+
|
107 |
+
# 清空历史,以免输入溢出
|
108 |
+
history = []
|
109 |
+
|
110 |
+
# 检测输入参数,如没有给定输入参数,直接退出
|
111 |
+
if os.path.exists(txt):
|
112 |
+
project_folder = txt
|
113 |
+
else:
|
114 |
+
if txt == "": txt = '空空如也的输入栏'
|
115 |
+
report_execption(chatbot, history, a=f"解析项目: {txt}", b=f"找不到本地项目或无权访问: {txt}")
|
116 |
+
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
117 |
+
return
|
118 |
+
|
119 |
+
# 搜索需要处理的文件清单
|
120 |
+
extensions = ['.mp4', '.m4a', '.wav', '.mpga', '.mpeg', '.mp3', '.avi', '.mkv', '.flac', '.aac']
|
121 |
+
|
122 |
+
if txt.endswith(tuple(extensions)):
|
123 |
+
file_manifest = [txt]
|
124 |
+
else:
|
125 |
+
file_manifest = []
|
126 |
+
for extension in extensions:
|
127 |
+
file_manifest.extend(glob.glob(f'{project_folder}/**/*{extension}', recursive=True))
|
128 |
+
|
129 |
+
# 如果没找到任何文件
|
130 |
+
if len(file_manifest) == 0:
|
131 |
+
report_execption(chatbot, history, a=f"解析项目: {txt}", b=f"找不到任何音频或视频文件: {txt}")
|
132 |
+
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
133 |
+
return
|
134 |
+
|
135 |
+
# 开始正式执行任务
|
136 |
+
yield from AnalyAudio(file_manifest, llm_kwargs, chatbot, history)
|
137 |
+
|
138 |
+
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|