小档案经过RIR，归一化后，合成大档案

关注

原创最新推荐文章于 2026-11-10 21:51:58 发布 1.4k 阅读 AI 写同款· GEO 优化 ›

[Web3] 基础第5篇: 共识算法与 P2P 通信协议 CometBFT 采用Tentmint 共识，一种易于理解、基本异步的 BFT 共识算法。该算法遵循一个简单的状态机，如下所示：上图展示了 Tendermint BFT 共识流程，描述了一个区块从提议（Propose）到最终提交（Commit）的完整过程。Tendermint 共识采用三阶段投票机制（Prevote、Precommit、Commit）来确保区块最终性。三阶段投票机制（Prevote → Precommit → Commit）需要 2/3 以上的投票支持才能进入下一阶段。阅读详情

import random
import os
import glob
import numpy as np
import scipy.io.wavfile as wav
import soundfile as sf
import librosa
from scipy import signal

import scipy.io as scio
dataFile = 'rir.mat'
rir = scio.loadmat(dataFile)

def merge_files(path_read_folder, path_write_wav_file):
#
files = os.listdir(path_read_folder)
merged_signal = []
cnt = 0
# for filename in glob.glob(os.path.join(path_read_folder, '*.wav')):
for filename in glob.glob(os.path.join(path_read_folder, '*.wav')):
clean,sr = librosa.load(filename, sr=32000)
ran = random.randint(0,2999)
reverb_speech = signal.fftconvolve(clean, rir['h'][ran,:], mode="full")
clean = clean/ max(abs(clean)) *32000 #* random.randint(4500,32000)
merged_signal.append(clean)

merged_signal=np.hstack(merged_signal)
merged_signal = np.asarray(merged_signal, dtype=np.int16)
wav.write(path_write_wav_file, 32000, merged_signal)

# for i in range (1):
# path_read_folder = "/Users/anker/Documents/1_AI/99_RIR/noise"
# path_write_wav_file = "/Users/anker/Documents/1_AI/99_RIR/noise" + str(i) + ".wav"
# merge_files(path_read_folder, path_write_wav_file)

path_read_folder = "/Volumes/Speech/语料/chinese1/"
path_write_wav_file = "/Volumes/Speech/语料/chinese1.wav"
merge_files(path_read_folder, path_write_wav_file)

收录于