148 lines
5.1 KiB
Python
148 lines
5.1 KiB
Python
from flask import Flask, request, send_file, jsonify
|
||
import requests
|
||
import os
|
||
import uuid
|
||
from datetime import datetime
|
||
from pydub import AudioSegment
|
||
from concurrent.futures import ThreadPoolExecutor
|
||
from queue import Queue
|
||
import json
|
||
from io import BytesIO
|
||
from chatgpt_api_config import chatgpt_apis
|
||
|
||
app = Flask(__name__)
|
||
|
||
tts_servers = [
|
||
'http://127.0.0.1:9995/tts',
|
||
'http://127.0.0.1:9996/tts'
|
||
]
|
||
tts_server_index = 0
|
||
executor = ThreadPoolExecutor(max_workers=len(tts_servers))
|
||
|
||
zh_punc = {'。', '？', '！', '\n'}
|
||
en_punc = {'.', '?', '!', '\n'}
|
||
|
||
def merge_audio_files(base_audio, increment):
|
||
"""将多段语音拼接"""
|
||
base_audio += increment
|
||
return base_audio
|
||
|
||
def call_tts_api(server_url, response_text, language, audio):
|
||
"""调用ChatTTS API，回答转语音"""
|
||
response = requests.post(
|
||
server_url,
|
||
data={
|
||
"text": response_text,
|
||
'language': language
|
||
},
|
||
files={'audio': open(audio, 'rb')}
|
||
)
|
||
if response.status_code == 200:
|
||
audio_segment = AudioSegment.from_file(file=BytesIO(response.content), format='wav')
|
||
return audio_segment
|
||
else:
|
||
print(f"Error: {response.json()['error']}")
|
||
return None
|
||
|
||
|
||
def generate_response_stream(transcription):
|
||
"""调用ChatGPT API，回答问题"""
|
||
for index, chatgpt_api in enumerate(chatgpt_apis):
|
||
url = chatgpt_api['url']
|
||
api_key = chatgpt_api['key']
|
||
|
||
headers = {
|
||
"Authorization": f"Bearer {api_key}",
|
||
"Content-Type": "application/json"
|
||
}
|
||
data = {
|
||
"model": "gpt-3.5-turbo",
|
||
"messages": [
|
||
{"role": "user", "content": transcription}
|
||
],
|
||
"temperature": 0.7,
|
||
"stream": True
|
||
}
|
||
response = requests.post(url, headers=headers, json=data, stream=True)
|
||
print(f"ChatGPT API {index} Response Status Code: {response.status_code}")
|
||
if response.status_code == 200:
|
||
return response
|
||
return None
|
||
|
||
|
||
@app.route('/tts', methods=['POST'])
|
||
def tts():
|
||
global tts_server_index
|
||
|
||
unique_id = str(uuid.uuid4())
|
||
timestamp = datetime.now().strftime('%Y%m%d%H%M%S')
|
||
os.makedirs('temp', exist_ok=True)
|
||
input_audio_filename = f"input_{timestamp}_{unique_id}.wav"
|
||
input_audio_path = os.path.join('temp', input_audio_filename)
|
||
output_audio_filename = f"output_{timestamp}_{unique_id}.wav"
|
||
output_audio_path = os.path.join('temp', output_audio_filename)
|
||
|
||
base_audio = AudioSegment.silent(duration=0) # 初始化一个空音频段
|
||
|
||
collected_chunks = []
|
||
collected_messages = ['']
|
||
futures = []
|
||
audio_queue = Queue()
|
||
|
||
language = request.form['language']
|
||
response_stream = generate_response_stream(request.form['text'])
|
||
if response_stream == None:
|
||
return jsonify({"error": "Something wrong with ChatGPT API."}), 502
|
||
speaker_file = request.files['audio']
|
||
speaker_file.save(input_audio_path)
|
||
|
||
try:
|
||
for chunk in response_stream.iter_lines():
|
||
if chunk:
|
||
decoded_line = chunk.decode('utf-8')
|
||
if decoded_line.startswith('data: '):
|
||
content = decoded_line[6:]
|
||
if content.strip() == '[DONE]':
|
||
break
|
||
response_json = json.loads(content)
|
||
collected_chunks.append(response_json)
|
||
chunk_message = response_json['choices'][0]['delta']
|
||
collected_messages[-1] += chunk_message.get('content', '')
|
||
|
||
if len(collected_messages[-1]) > 0 and collected_messages[-1][-1] in (zh_punc if language == 'chinese' else en_punc):
|
||
partial_text = collected_messages[-1]
|
||
if partial_text:
|
||
print(f"{partial_text}", end="")
|
||
server_url = tts_servers[tts_server_index % len(tts_servers)]
|
||
tts_server_index += 1
|
||
future = executor.submit(call_tts_api, server_url, partial_text, language, input_audio_path)
|
||
futures.append((partial_text, future))
|
||
collected_messages.append("")
|
||
|
||
# 处理所有 future 并按顺序添加到队列中
|
||
for partial_text, future in futures:
|
||
audio_data = future.result()
|
||
if audio_data:
|
||
audio_queue.put((partial_text, audio_data))
|
||
|
||
# 拼接音频文件
|
||
while not audio_queue.empty():
|
||
_, audio_segment = audio_queue.get()
|
||
base_audio = merge_audio_files(base_audio, audio_segment)
|
||
|
||
# 将最终的音频文件保存到硬盘
|
||
base_audio.export(output_audio_path, format='wav')
|
||
print("\n")
|
||
|
||
# 返回生成的回答音频
|
||
return send_file(output_audio_path, as_attachment=True, download_name='response.wav')
|
||
finally:
|
||
if os.path.exists(input_audio_path):
|
||
os.remove(input_audio_path)
|
||
if os.path.exists(output_audio_path):
|
||
os.remove(output_audio_path)
|
||
|
||
|
||
if __name__ == '__main__':
|
||
app.run()
|