lanhao
53 lines
1.9 KiB
Python
53 lines
1.9 KiB
Python
import os
|
|
import requests
|
|
|
|
def synthesize_cosyvoice(text, speed=1.0, pitch=0, volume=1.0, instruction=None):
|
|
"""调用阿里云百炼CosyVoice API"""
|
|
api_key = os.getenv('DASHSCOPE_API_KEY', '')
|
|
if not api_key:
|
|
raise ValueError('DASHSCOPE_API_KEY未设置')
|
|
|
|
url = 'https://dashscope.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer'
|
|
headers = {
|
|
'Authorization': f'Bearer {api_key}',
|
|
'Content-Type': 'application/json'
|
|
}
|
|
|
|
# 前端 pitch: -12~12 → API rate: 0.5~2.0（以1.0为中心）
|
|
pitch_ratio = max(0.5, min(2.0, 1.0 + pitch / 24))
|
|
# 前端 volume: 0~10 → API volume: 0~100
|
|
volume_int = int(max(0, min(100, volume * 10)))
|
|
|
|
payload = {
|
|
'model': 'cosyvoice-v3.5-flash',
|
|
'input': {
|
|
'text': text,
|
|
'voice': 'cosyvoice-v3.5-flash-bailian-e859a6e87823419f9c6b3eefb1a97dc3',
|
|
'format': 'mp3',
|
|
'sample_rate': 32000
|
|
},
|
|
'parameters': {
|
|
'rate': max(0.5, min(2.0, speed)),
|
|
'pitch': pitch_ratio,
|
|
'volume': volume_int
|
|
}
|
|
}
|
|
|
|
response = requests.post(url, json=payload, headers=headers, timeout=60)
|
|
|
|
if response.status_code != 200:
|
|
raise Exception(f'CosyVoice API错误: {response.status_code} - {response.text}')
|
|
|
|
result = response.json()
|
|
|
|
# 阿里云返回JSON格式，音频数据在 output.audio.url 字段
|
|
if 'output' in result and 'audio' in result['output']:
|
|
audio_url = result['output']['audio'].get('url')
|
|
if audio_url:
|
|
# 从URL下载实际音频
|
|
audio_response = requests.get(audio_url, timeout=60)
|
|
if audio_response.status_code == 200:
|
|
return audio_response.content
|
|
raise Exception(f'下载音频失败: {audio_response.status_code}')
|
|
|
|
raise Exception(f'CosyVoice返回格式异常: {result}') |