mirror of
https://github.com/HiMeditator/auto-caption.git
synced 2026-02-04 12:24:42 +08:00
109 lines
3.7 KiB
Python
109 lines
3.7 KiB
Python
import samplerate
|
|
import numpy as np
|
|
import numpy.core.multiarray # do not remove
|
|
|
|
def merge_chunk_channels(chunk: bytes, channels: int) -> bytes:
|
|
"""
|
|
将当前多通道音频数据块转换为单通道音频数据块
|
|
|
|
Args:
|
|
chunk: 多通道音频数据块
|
|
channels: 通道数
|
|
|
|
Returns:
|
|
单通道音频数据块
|
|
"""
|
|
if channels == 1: return chunk
|
|
# (length * channels,)
|
|
chunk_np = np.frombuffer(chunk, dtype=np.int16)
|
|
# (length, channels)
|
|
chunk_np = chunk_np.reshape(-1, channels)
|
|
# (length,)
|
|
chunk_mono_f = np.mean(chunk_np.astype(np.float32), axis=1)
|
|
chunk_mono = np.round(chunk_mono_f).astype(np.int16)
|
|
return chunk_mono.tobytes()
|
|
|
|
|
|
def resample_chunk_mono(chunk: bytes, channels: int, orig_sr: int, target_sr: int, mode="sinc_best") -> bytes:
|
|
"""
|
|
将当前多通道音频数据块转换成单通道音频数据块，然后进行重采样
|
|
|
|
Args:
|
|
chunk: 多通道音频数据块
|
|
channels: 通道数
|
|
orig_sr: 原始采样率
|
|
target_sr: 目标采样率
|
|
mode: 重采样模式，可选：'sinc_best' | 'sinc_medium' | 'sinc_fastest' | 'zero_order_hold' | 'linear'
|
|
|
|
Return:
|
|
单通道音频数据块
|
|
"""
|
|
if channels == 1:
|
|
chunk_mono = np.frombuffer(chunk, dtype=np.int16)
|
|
chunk_mono = chunk_mono.astype(np.float32)
|
|
else:
|
|
# (length * channels,)
|
|
chunk_np = np.frombuffer(chunk, dtype=np.int16)
|
|
# (length, channels)
|
|
chunk_np = chunk_np.reshape(-1, channels)
|
|
# (length,)
|
|
chunk_mono = np.mean(chunk_np.astype(np.float32), axis=1)
|
|
|
|
ratio = target_sr / orig_sr
|
|
chunk_mono_r = samplerate.resample(chunk_mono, ratio, converter_type=mode)
|
|
chunk_mono_r = np.round(chunk_mono_r).astype(np.int16)
|
|
return chunk_mono_r.tobytes()
|
|
|
|
|
|
def resample_chunk_mono_np(chunk: bytes, channels: int, orig_sr: int, target_sr: int, mode="sinc_best", dtype=np.float32) -> np.ndarray:
|
|
"""
|
|
将当前多通道音频数据块转换成单通道音频数据块，然后进行重采样，返回 Numpy 数组
|
|
|
|
Args:
|
|
chunk: 多通道音频数据块
|
|
channels: 通道数
|
|
orig_sr: 原始采样率
|
|
target_sr: 目标采样率
|
|
mode: 重采样模式，可选：'sinc_best' | 'sinc_medium' | 'sinc_fastest' | 'zero_order_hold' | 'linear'
|
|
dtype: 返回 Numpy 数组的数据类型
|
|
|
|
Return:
|
|
单通道音频数据块
|
|
"""
|
|
if channels == 1:
|
|
chunk_mono = np.frombuffer(chunk, dtype=np.int16)
|
|
chunk_mono = chunk_mono.astype(np.float32)
|
|
else:
|
|
# (length * channels,)
|
|
chunk_np = np.frombuffer(chunk, dtype=np.int16)
|
|
# (length, channels)
|
|
chunk_np = chunk_np.reshape(-1, channels)
|
|
# (length,)
|
|
chunk_mono = np.mean(chunk_np.astype(np.float32), axis=1)
|
|
|
|
ratio = target_sr / orig_sr
|
|
chunk_mono_r = samplerate.resample(chunk_mono, ratio, converter_type=mode)
|
|
chunk_mono_r = chunk_mono_r.astype(dtype)
|
|
return chunk_mono_r
|
|
|
|
|
|
def resample_mono_chunk(chunk: bytes, orig_sr: int, target_sr: int, mode="sinc_best") -> bytes:
|
|
"""
|
|
将当前单通道音频块进行重采样
|
|
|
|
Args:
|
|
chunk: 单通道音频数据块
|
|
orig_sr: 原始采样率
|
|
target_sr: 目标采样率
|
|
mode: 重采样模式，可选：'sinc_best' | 'sinc_medium' | 'sinc_fastest' | 'zero_order_hold' | 'linear'
|
|
|
|
Return:
|
|
单通道音频数据块
|
|
"""
|
|
chunk_np = np.frombuffer(chunk, dtype=np.int16)
|
|
chunk_np = chunk_np.astype(np.float32)
|
|
ratio = target_sr / orig_sr
|
|
chunk_r = samplerate.resample(chunk_np, ratio, converter_type=mode)
|
|
chunk_r = np.round(chunk_r).astype(np.int16)
|
|
return chunk_r.tobytes()
|