[2noise/ChatTTS]找到一种固定音色的方法

具体思路： 1. 生成rand_spk并把weight保存到csv文件：

import torch
import csv
std, mean = torch.load("spk_stat.pt').chunk(2)

rand_spk = torch.randn(768) * std + mean
writeToCsv(f"saved.csv",rand_spk.detach().numpy())

def writeToCsv(csv_file_path,data):
  with open(csv_file_path, mode='w', newline='') as file:
    writer = csv.writer(file)
    # 写入数据
    writer.writerow(data.tolist())

2. 生成语音时加载保存的weight，注意temperature要设成一个极小的值

import pandas as pd

data = pd.read_csv(f"./saved.csv", header=None)
rand_spk = torch.tensor(data.iloc[0], dtype=torch.float32)
params_infer_code = {
    'spk_emb': rand_spk,  # add sampled speaker
    'temperature': .000000000001,  # using custom temperature
    'top_P': 0.7,  # top P decode
    'top_K': 20,  # top K decode
  }
params_refine_text = {
  'prompt': '[break_2]'
}
wavs = chat.infer("你的文本"
                      , params_refine_text=params_refine_text, params_infer_code=params_infer_code, use_decoder=True)
scipy.io.wavfile.write(filename=f"./chattts_download.wav", rate=24_000, data=wavs[0].T)