4 New Audio Time Stretching Algorithms

Under construction...

This page is a preview of 4 new audio time stretching algorithms currently in development.

The algorithms have the code-names "Quantum", "Tachyon", "Schrödinger" and "Heisenberg".

Audio examples
Performance comparison
Software
Scientific References
Musical References
History

Audio examples

A few audio examples are much better than a lot of words...

Notes:

Throughout this page, playing any audio excerpt will automatically stop the previous one if not finished (unless JavaScript is disabled)
Transformed excerpts are truncated to 30 s (60 s for 20x stretch)
All audio excerpts are loudness-normalized using EBU R128 to facilitate comparisons

Original	Electronic [1]
Time stretching ratio	Quantum	Tachyon	Schrödinger	Heisenberg
0.5
0.9
1.3
2.0
4.0
Cycle 2.0, 0.6, 1.5, 5/9
0.2
20

Original	Pop [2]
Time stretching ratio	Quantum	Tachyon	Schrödinger	Heisenberg
0.5
0.9
1.3
2.0
4.0
Cycle 2.0, 0.6, 1.5, 5/9
0.2
20

Original	Guitar [3]
Time stretching ratio	Quantum	Tachyon	Schrödinger	Heisenberg
0.5
0.9
1.3
2.0
4.0
Cycle 2.0, 0.6, 1.5, 5/9
0.2
20

Original	New Age [4]
Time stretching ratio	Quantum	Tachyon	Schrödinger	Heisenberg
0.5
0.9
1.3
2.0
4.0
Cycle 2.0, 0.6, 1.5, 5/9
0.2
20

High quality comes with a price, usually slow processing speed. The following table shows an approximation, for each algorithm, of the time required to produce one minute of time stretched audio (stereo, 44.1 kHz) on a 3 GHz processor. It also shows the relative quality (subjective, your opinion might differ), the minimum and maximum supported time stretching ratios, and the main goals:

Time to produce 1 min Quality (Subjective) Transformation ratios
Max / Recommanded Main Goals

Quantum 5 - 6 seconds 😐 Medium 😐 0.25 - 4 / 0.7 - 1.3 Best speed/quality trade-off, simplicity

Tachyon 10 - 15 minutes 😱 Highest 😃 0.25 - 4 / 0.5 - 2 Highest quality at any price

Schrödinger < 2 seconds 😃 Lowest 😒 0.25 - 4 / 0.8 - 1.2 Fastest speed, for solo, open source

Heisenberg ~ 30 seconds 😒 Medium 😐 0.01 - 1000 / 0.1 - 100 Huge transformation ratios, pleasant sound

	Time to produce 1 min	Quality (Subjective)	Transformation ratios Max / Recommanded	Main Goals
Quantum	5 - 6 seconds	😐	Medium	😐	0.25 - 4 / 0.7 - 1.3	Best speed/quality trade-off, simplicity
Tachyon	10 - 15 minutes	😱	Highest	😃	0.25 - 4 / 0.5 - 2	Highest quality at any price
Schrödinger	< 2 seconds	😃	Lowest	😒	0.25 - 4 / 0.8 - 1.2	Fastest speed, for solo, open source
Heisenberg	~ 30 seconds	😒	Medium	😐	0.01 - 1000 / 0.1 - 100	Huge transformation ratios, pleasant sound

Software

The algorithms are all proptotypes. Development is in progress. For the moment only the Quantum algorithm is available as a command-line, beta version.

This page will be updated as progress is made...

Download

Download TimeStretchQuantum.7z, a beta-version of the Quantum audio time stretching algorithm, as a command-line tool.

Usage

java -jar TimeStretchQuantum.jar <input file> <time-stretch ratio> <output-file>

Limitations

You need to install a Java Runtime Environment version 7 or greater.
The tool can only be used from the command line. There is no graphical user interface
The tool can only save in the .wav, .aiff or .au audio format, and the supported input formats are limited.
The tool will not prevent clipping, which is frequent when time stretching, especially with modern masterings. You may need to reduce the gain manually before time stretching.

Scientific References

The science behind these algorithms...

[1] Zdeněk Průša and Nicki Holighaus, "Phase Vocoder Done Right", 25^th European Signal Processing Conference, EUSIPCO, 2017 ^{(1), (2)}

[2] N. Juillerat, "Audio Time Stretching with Controllable Phase Coherence", 142^nd Audio Engineering Society Convention, AES 142, 2017 ⁽²⁾

[3] N. Juillerat, B. Hirsbrunner, "Audio Time Stretching with an Adaptive Multiresolution Phase Vocoder", IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2017 ^{(2), (4)}

[4] Derry FitzGerald, "Harmonic/Percussive Separation using Median Filtering", proc. of the 13^th intl. conf on Digital Audio Effects, 2010 ⁽³⁾

[5] N. Juillerat, S. Schubiger-Banz, S. M. Arisona, "Enhancing the Quality of Audio Transformations Using the Multi-Scale Short-Time Fourier Transform", IASTED International Conference on Signal Processing, IASTED SIP 2008 ^{(2), (4)}

[6] Tan Roland K. C. and Lin Amerson H. J, "A Time-Scale Modification Algorithm Based on the Subband Time-Domain Technique for Broad-Band Signal Applications", Journal of the Audio Engineering Society, vol. 48, no. 5, pp. 437-449, 2000 ⁽³⁾

[7] Jean Laroche and Mark Dolson, "Improved Phase Vocoder Time-Scale Modification of Audio", IEEE Transactions on Speech and Signal Processing, vol. 7, no. 3, May 1999 ⁽⁴⁾

[8] Werner Verhelst, Marc Roelands, "An Overlap-Add Technique Based on Waveform Similarity (WSOLA) for High Quality Time-Scale Modification of Speech", IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1993 ⁽³⁾

Notes:

Used partly and among others in ⁽¹⁾Quantum, ⁽²⁾Tachyon, ⁽³⁾Schrödinger, ⁽⁴⁾Heisenberg

Musical References

[1] Jean-Michel Jarre, "Oxygène part. 4", Oxygène, Disques Dreyfus, 1977

[2] Peter Gabriel, "Sledgehammer", Secret World Live, Geffen Records, 1986

[3] (Unknown) Sample n° 4 (Guitare) from "Comparatif Time Stretching et Pitch Shifting", Audiofanzine, 2008

[4] Melanie G 'MySweetDarkness', "The Morn", Through the Shadow and the Light, Everlasting Dream, 2007

History

2019-07-05: First version

PitchTech home page