soxr package#

class soxr.ResampleStream(in_rate: float, out_rate: float, num_channels: int, dtype='float32', quality='HQ')#

Streaming resampler

Use ResampleStream for real-time processing or very long signal.

Parameters:

in_rate (float) – Input sample-rate.
out_rate (float) – Output sample-rate.
num_channels (int) – Number of channels.
dtype (type or str, optional) – Internal data type processed with. Should be one of float32, float64, int16, int32.
quality (int or str, optional) – Quality setting. One of QQ, LQ, MQ, HQ, VHQ.

clear() → None#

Reset resampler. Ready for fresh signal, same config.

This can be used to save initialization time.

delay() → float#

Get current delay.

SoXR output has an algorithmic delay. This function returns the length of current pending output.

num_clips() → int#

Clip counter. (for int I/O)

resample_chunk(x: numpy.ndarray, last=False) → numpy.ndarray#

Resample chunk with streaming resampler

Parameters:

x (np.ndarray) – Input array. Input can be mono(1D) or multi-channel(2D of [frame, channel]). dtype should match with constructor.
last (bool, optional) – Set True at final chunk to flush last outputs. It should be True only once at the end of a continuous sequence.

Returns:

Resampled data. Output is np.ndarray with same ndim with input.

Return type:

np.ndarray

soxr.resample(x: numpy.typing.ArrayLike, in_rate: float, out_rate: float, quality='HQ') → numpy.ndarray#

Resample signal

Parameters:

x (array_like) – Input array. Input can be mono(1D) or multi-channel(2D of [frame, channel]). If input is not np.ndarray, it will be converted to np.ndarray(dtype=’float32’). Its dtype should be one of float32, float64, int16, int32.
in_rate (float) – Input sample-rate.
out_rate (float) – Output sample-rate.
quality (int or str, optional) – Quality setting. One of QQ, LQ, MQ, HQ, VHQ.

Returns:

Resampled data. Output is np.ndarray with same ndim and dtype with input.

Return type:

np.ndarray

soxr package