timeStretch: Time stretch
In soundgen: Sound Synthesis and Acoustic Analysis

timeStretch

R Documentation

Time stretch

Description

Dynamically time-stretches a sound without preserving its pitch or formants, as if gradually changing playback speed. Algorithm: the audio is resampled at time-varying steps. This is about 100 times faster than time-stretching with a phase vocoder in shiftPitch, but pitch and formants cannot be preserved, and compressing the sound (stretch < 1) may cause artifacts due to aliasing, since no anti-aliasing filter is applied (low-pass filter the input and/or output if needed). The stretch contour is interpreted over the output (stretched) duration, which is why the trajectory differs from shiftPitch with the same anchors.

Usage

timeStretch(
  x,
  stretch = 1,
  samplingRate = NULL,
  interpol = "splineFC",
  precision = 1000,
  play = FALSE,
  saveAudio = FALSE,
  reportEvery = NULL,
  cores = 1
)

Arguments

`x`	path to a folder, one or more wav or mp3 files c('file1.wav', 'file2.mp3'), Wave object, numeric vector, or a list of Wave objects or numeric vectors
`stretch`	1 = no change, >1 = longer, <1 = shorter. Single value, vector, or anchor format (see `soundgen`). Must be positive
`samplingRate`	sampling rate of `x` (only needed if `x` is a numeric vector)
`interpol`	interpolation method supported by `interpolate`
`precision`	the number of points used for estimating the duration of output (more = better, but slower)
`play`	if TRUE, plays the output audio using the default player on your system. If a character string, it is passed to `playme` as the name of the player to use (e.g. 'aplay', 'play', 'vlc'). In case of errors, try setting another default player for `playme`
`saveAudio`	if TRUE, saves the processed audio in a subdirectory named after the function and created in the input directory (if input is a file or folder) or in the working directory
`reportEvery`	when processing multiple inputs, report estimated time left every `reportEvery` iterations (NULL = default, NA = don't report); see `reportTime`
`cores`	number of cores for parallel processing

Value

The processed waveform as a numeric vector for a single input or a list for multiple outputs. The duration is approximate: the output may end up a few samples shorter than intended if the resampling trajectory reaches the end of the input.

Examples

data(speechEx, package = 'soundgen')  # import a recording
# playme(speechEx)
# spectrogram(speechEx)
s1 = timeStretch(speechEx, stretch = c(1, 3))
# playme(s1, speechEx@samp.rate)
# spectrogram(s1, speechEx@samp.rate)

# compare to a similar effect achieved with a phase vocoder in shiftPitch():
s2 = shiftPitch(
  speechEx,
  timeStretch = c(1, 3),  # from 1 (original) to mult
  multPitch = c(1, 1/3),  # also drop pitch
  multFormants = c(1, 1/3)  # also drop formants (by the same proportion)
)
# playme(s2, speechEx@samp.rate)
# spectrogram(s2, speechEx@samp.rate)
# NB: because the two algorithms calculate transitions between stretch
# factors in different ways, the duration is not identical, even though the
# range of pitch change is the same

soundgen documentation built on Sept. 20, 2026, 5:07 p.m.