-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain.py
249 lines (220 loc) · 8.29 KB
/
main.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
import gradio as gr
import requests
import llama_podcast
import openai
import json
import soundfile as sf
import numpy as np
import os
opt = gr.WaveformOptions()
opt.sample_rate = 16000
audios = {}
tts_inputs = []
if not os.path.exists("./output"):
os.makedirs("./output")
def set_tts_inputs(inputs: list):
tts_inputs.clear()
tts_inputs.extend(inputs)
# TTS 功能
def tts(index, base_url, speaker, text):
# 发起HTTP请求到后端，获取wav文件
response = requests.post(
base_url,
json={"speaker": speaker, "input": text},
)
# 确保请求成功
if response.status_code == 200:
file_path = f"./output/segments_{index}.wav"
with open(file_path, "wb") as file:
file.write(response.content)
audios[index] = file_path
return file_path
else:
return None
def seq_tts(base_url, speaker1, speaker2):
size = len(tts_inputs)
for i, text in enumerate(tts_inputs):
speaker, text = text
if speaker == "Speaker 1":
audio = tts(str(i), base_url, speaker1, text)
else:
audio = tts(str(i), base_url, speaker2, text)
yield f"完成 {i+1}/{size}"
return "合成完成"
def merge_audio():
wav_data = []
for i in range(len(tts_inputs)):
audio = audios[str(i)]
data, fs = sf.read(audio)
wav_data.append(data)
wav_data = np.concatenate(wav_data)
SAMPLE_RATE = 32000
sf.write("./output/podcast.wav", wav_data, SAMPLE_RATE)
return "./output/podcast.wav"
lang = os.environ.get("LANG", "zh")
if lang == "zh":
default_sys_prompt1 = llama_podcast.CN_SYSTEMP_PROMPT_1
default_sys_prompt2 = llama_podcast.CN_SYSTEMP_PROMPT_2
else:
default_sys_prompt1 = llama_podcast.EN_SYSTEMP_PROMPT_1
default_sys_prompt2 = llama_podcast.EN_SYSTEMP_PROMPT_2
with gr.Blocks() as demo:
with gr.Column():
llm_base_url = gr.Textbox(
label="LLM BaseURL",
value="https://llama70b.gaia.domains/v1 ",
)
llm_model_name = gr.Textbox(
label="Model Name",
value="llama",
)
llm_token = gr.Textbox(
label="API Token",
value="NA",
)
with gr.Tab("1. 撰写演讲稿"):
llm_sys_prompt = gr.TextArea(
label="System Prompt",
value=default_sys_prompt1,
)
input_prompt = gr.TextArea(label="文本内容")
button = gr.Button(value="开始生成")
output1 = gr.TextArea(label="LLM Output", value="")
@gr.on(
triggers=[button.click],
inputs=[
llm_base_url,
llm_token,
llm_model_name,
llm_sys_prompt,
input_prompt,
],
outputs=[output1],
)
def generate_llm(
llm_base_url: str,
llm_token: str,
llm_model_name: str,
llm_sys_prompt,
input_prompt,
):
openapi_client = openai.Client(
base_url=llm_base_url.strip(),
api_key=llm_token.strip(),
timeout=1000 * 60 * 60,
)
messages = [
{"role": "system", "content": llm_sys_prompt},
{"role": "user", "content": input_prompt},
]
response = openapi_client.chat.completions.create(
messages=messages, model=llm_model_name, stream=True
)
outputs = ""
for chunk in response:
outputs += chunk.choices[0].delta.content
yield outputs
with gr.Tab("2. 润色演讲稿"):
llm_sys_prompt = gr.TextArea(
label="System Prompt",
value=default_sys_prompt2,
)
input_prompt = gr.TextArea(label="文本内容", value=output1.value)
update_btn = gr.Button(value="刷新演讲稿")
button = gr.Button(value="开始生成")
output2 = gr.TextArea(label="LLM Output")
update_btn.click(lambda x: x, inputs=[output1], outputs=[input_prompt])
@gr.on(
triggers=[button.click],
inputs=[
llm_base_url,
llm_token,
llm_model_name,
llm_sys_prompt,
input_prompt,
],
outputs=[output2],
)
def generate_llm(
llm_base_url,
llm_token,
llm_model_name,
llm_sys_prompt,
input_prompt,
):
openapi_client = openai.Client(
base_url=llm_base_url.strip(),
api_key=llm_token.strip(),
timeout=1000 * 60 * 60,
)
messages = [
{"role": "system", "content": llm_sys_prompt},
{"role": "user", "content": input_prompt},
]
response = openapi_client.chat.completions.create(
messages=messages, model=llm_model_name, stream=True
)
outputs = ""
for chunk in response:
outputs += chunk.choices[0].delta.content
yield outputs
with gr.Tab("3. 生成播客 TTS"):
tts_base_url = gr.Textbox(
label="TTS BaseURL",
value="http://localhost:8080/v1/audio/speech_gpt",
)
if lang == "zh":
speaker1 = gr.Textbox(label="Speaker1", value="cctv_male_anchor")
speaker2 = gr.Textbox(label="Speaker2", value="cctv_female_anchor")
else:
speaker1 = gr.Textbox(label="Speaker1", value="cooper")
speaker2 = gr.Textbox(label="Speaker2", value="kelly")
text_input = gr.TextArea(label="演讲稿", value="")
update_btn = gr.Button(value="刷新演讲稿")
update_btn.click(lambda x: x, inputs=[output2], outputs=[text_input])
seq_tts_btn = gr.Button(value="逐句合成")
label = gr.Label("未合成音频")
seq_tts_btn.click(
seq_tts,
inputs=[tts_base_url, speaker1, speaker2],
outputs=[label],
)
@gr.render(inputs=[text_input, label])
def update_tts_input(input, _label):
with gr.Column("tts_inputs") as col:
try:
texts = json.loads(input)
except:
texts = []
set_tts_inputs(texts)
for i, text in enumerate(texts):
speaker, text = text
with gr.Row():
index = gr.Textbox(label="Index", value=str(i))
text = gr.Textbox(label=speaker, value=text)
with gr.Row():
if str(i) in audios:
audio = gr.Audio(value=audios[str(i)], type="filepath")
else:
audio = gr.Audio()
btn = gr.Button(value="重新生成")
if speaker == "Speaker 1":
btn.click(
tts,
inputs=[index, tts_base_url, speaker1, text],
outputs=[audio],
)
else:
btn.click(
tts,
inputs=[index, tts_base_url, speaker2, text],
outputs=[audio],
)
return col
gr.Label("合并音频")
final_audio = gr.Audio(type="filepath")
button = gr.Button(value="合并")
button.click(merge_audio, outputs=[final_audio])
# 启动应用
if __name__ == "__main__":
demo.launch()