qqbot/tts.py

import csv
import datetime
import os
import requests
from config import key_get
import base64

free_map = {
    "诺艾尔": "原神-中文-诺艾尔_ZH",
    "芙宁娜": "原神-中文-芙宁娜_ZH",
    "神里绫华": "原神-中文-神里绫华_ZH",
    "迪卢克": "原神-中文-迪卢克_ZH",
}
def get_free_tts(name, text):
    """
    从免费的试用api获取语音合成效果
    """
    url = "https://gsv2p.acgnai.top/infer_single"
    headers = {
        "Authorization": "Bearer " + key_get('tts'),
        "Content-Type": "application/json"
    }
    model = free_map.get(name)
    if model is None:
        return "该模型暂不支持"
    response = requests.post(url, headers=headers, json={
          "version": "v4",
          "model_name": model,
          "prompt_text_lang": "中文",
          "emotion": "默认",
          "text": text,
          "text_lang": "中文",
          "top_k": 10,
          "top_p": 1,
          "temperature": 1,
          "text_split_method": "按标点符号切",
          "batch_size": 1,
          "batch_threshold": 0.75,
          "split_bucket": True,
          "speed_facter": 1,
          "fragment_interval": 0.3,
          "media_type": "mp3",
          "parallel_infer": True,
          "repetition_penalty": 1.35,
          "seed": -1,
          "sample_steps": 16,
          "if_sr": False
        })
    data = response.json()
    if data.get("msg") == "合成成功":
        return data.get("audio_url")
    else:
        return "获取失败"

def get_gpu_tts(text, text_lang="ja",
                file_path="/home/gxx/git/GPT-SoVITS-main/vioce/Future Samples_Anime Vocals Vol. 2_Vocal 35.wav",
                prompt="お願い、もう少しここにいてもいいかな"):
    """
    使用gpu进行v2推理，默认霓虹文
    """
    url = "http://127.0.0.1:9880/tts"
    # url = "https://api.gxx12138.space/vioce/tts"
    paload = {
        "text": text,
        "text_lang": text_lang,
        "ref_audio_path": file_path,
        "aux_ref_audio_paths": [],
        "prompt_text": prompt,
        "prompt_lang": text_lang,
        "top_k": 5,
        "top_p": 1,
        "temperature": 1,
        "text_split_method": "cut0",
        "batch_size": 1,
        "batch_threshold": 0.75,
        "split_bucket": True,
        "speed_factor":1.0,
        "streaming_mode": False,
        "seed": -1,
        "parallel_infer": True,
        "repetition_penalty": 1.35,
        "sample_steps": 32,
        "super_sampling": False
    }
    # api返回wav文件，转为base64编码返回
    response = requests.post(url, json=paload)
    if response.status_code != 200:
        return "获取失败"
    bs64 = base64.b64encode(response.content).decode('utf-8')
    return bs64


if __name__ == "__main__":
    print(get_gpu_tts("おやすみにゃ～。みんな素敵な夢を見られますようににゃ。"))