Caution

You're reading an old version of this documentation. If you want up-to-date information, please have a look at 0.10.2.

librosa.resample

librosa.resample(y, *, orig_sr, target_sr, res_type='soxr_hq', fix=True, scale=False, axis=-1, **kwargs)[source]

Resample a time series from orig_sr to target_sr

By default, this uses a high-quality method (soxr_hq) for band-limited sinc interpolation. The alternate res_type values listed below offer different trade-offs of speed and quality.

Parameters:

ynp.ndarray [shape=(…, n, …)]

audio time series, with n samples along the specified axis.

orig_srnumber > 0 [scalar]

original sampling rate of y

target_srnumber > 0 [scalar]

target sampling rate

res_typestr (default: soxr_hq)

resample type

‘soxr_vhq’, ‘soxr_hq’, ‘soxr_mq’ or ‘soxr_lq’: soxr Very high-, High-, Medium-, Low-quality FFT-based bandlimited interpolation. 'soxr_hq' is the default setting of soxr.
‘soxr_qq’: soxr Quick cubic interpolation (very fast, but not bandlimited)
‘kaiser_best’: resampy high-quality mode
‘kaiser_fast’: resampy faster method
‘fft’ or ‘scipy’: scipy.signal.resample Fourier method.
‘polyphase’: scipy.signal.resample_poly polyphase filtering. (fast)
‘linear’: samplerate linear interpolation. (very fast, but not bandlimited)
‘zero_order_hold’: samplerate repeat the last value between samples. (very fast, but not bandlimited)
‘sinc_best’, ‘sinc_medium’ or ‘sinc_fastest’: samplerate high-, medium-, and low-quality bandlimited sinc interpolation.

Note

Not all options yield a bandlimited interpolator. If you use soxr_qq, polyphase, linear, or zero_order_hold, you need to be aware of possible aliasing effects.

Note

samplerate and resampy are not installed with librosa. To use samplerate or resampy, they should be installed manually:

$ pip install samplerate
$ pip install resampy

Note

When using res_type='polyphase', only integer sampling rates are supported.

fixbool

adjust the length of the resampled signal to be of size exactly ceil(target_sr * len(y) / orig_sr)

scalebool

Scale the resampled signal so that y and y_hat have approximately equal total energy.

axisint

The target axis along which to resample. Defaults to the trailing axis.

**kwargsadditional keyword arguments

If fix==True, additional keyword arguments to pass to librosa.util.fix_length.

Returns:

y_hatnp.ndarray [shape=(…, n * target_sr / orig_sr, …)]: y resampled from orig_sr to target_sr along the target axis

Raises:

ParameterError: If res_type='polyphase' and orig_sr or target_sr are not both integer-valued.