如何更快地将大量 PCM 样本编码/解码为 IMA ADPCM 样本？

问题描述

我想尽快将 wav 文件编码为 IMA ADPCM。
但是当我尝试编码我的文件（1:​​39 长度，44100 采样率，16 位，立体声）时，编码它需要大约 8 秒...
有没有办法让它更快？因为 8 秒对我来说看起来很慢...
另外，我尝试使用 Python 的 audioop 模块，但它仍然很慢，而且我真的不需要 audioop 的东西，所以我编写了自己的示例编码器/解码器。

我的代码：

steps = [
    7,8,9,10,11,12,13,14,16,17,19,21,23,25,28,31,34,37,41,45,50,55,60,66,73,80,88,97,107,118,130,143,157,173,190,209,230,253,279,307,337,371,408,449,494,544,598,658,724,796,876,963,1060,1166,1282,1411,1552,1707,1878,2066,2272,2499,2749,3024,3327,3660,4026,4428,4871,5358,5894,6484,7132,7845,8630,9493,10442,11487,12635,13899,15289,16818,18500,20350,22385,24623,27086,29794,32767
]

step_indices = [-1,-1,2,4,6,8]

def clamp(value,lower,upper):
    return lower if (value < lower) else upper if (value > upper) else value

def decode_sample(sample,state: tuple) -> tuple:
    predicted_sample = state[0] if (state) else 0
    step_index = state[1] if (state) else 0
    step = steps[step_index]

    diff = step >> 3
    if sample & 1: diff += step >> 2
    if sample & 2: diff += step >> 1
    if sample & 4: diff += step
    if sample & 8: diff = -diff
    
    predicted_sample = clamp(predicted_sample + diff,-32768,32767)
    step_index = clamp(step_index + step_indices[sample & 7],88)

    return predicted_sample,step_index

def encode_sample(sample,state: tuple) -> tuple:
    predicted_sample = state[0] if (state) else 0
    step_index = state[1] if (state) else 0
    step = steps[step_index]

    sample_diff = sample - predicted_sample
    encoded_sample = 8 if (sample_diff < 0) else 0

    if encoded_sample:
        sample_diff = -sample_diff
    
    diff = step >> 3
    if sample_diff >= step:
        encoded_sample |= 4
        sample_diff -= step
        diff += step

    step >>= 1
    if sample_diff >= step:
        encoded_sample |= 2
        sample_diff -= step
        diff += step

    step >>= 1
    if sample_diff >= step:
        encoded_sample |= 1
        diff += step

    if encoded_sample & 8:
        diff = -diff
    
    predicted_sample = clamp(predicted_sample + diff,32767)
    step_index = clamp(step_index + step_indices[encoded_sample & 7],88)

    return encoded_sample,(predicted_sample,step_index)

import wave

def encode_stereo():
    with wave.open('mario.wav','r') as wav_file:
        raw_samples = wav_file.readframes(wav_file.getnframes())
        samples = memoryview(raw_samples).cast('h')
        left_channel,right_channel = samples[::2],samples[1::2]

        encoded = bytearray()
        left_state = None
        right_state = None
        
        for i in range(len(left_channel)):
            left_sample,left_state = encode_sample(left_channel[i],left_state)
            right_sample,right_state = encode_sample(right_channel[i],right_state)
            
            encoded.append(right_sample << 4 | left_sample)

        with open('mario.adpcm','wb') as adpcm_file:
            adpcm_file.write(encoded)

encode_stereo()

解决方法

我想知道您是否见过 pyima - 它似乎在做同样的事情。

如果要加快速度 - 您可以并行运行几个文件的编码 - 如果批量工作，这是您可以做的最简单的事情。但是对于单个文件编码 - 我建议使用一些优化良好的 C++ 库，它具有 Python 接口来进行编码。