Two Questions about SBSMS

Dear developers and users,

  1. Is the SBSMS (Subband Sinusoidal Modeling Synthesis) a kind of ‘overlap add time stretching’?
    https://www.google.co.uk/books/edition/Introduction_to_Computer_Music/mCpqmESEOEcC?hl=en&gbpv=1&dq=‘overlap+add+time+stretching’+computer+music+nick+collins&pg=PA150&printsec=frontcover

  2. Does Audacity resample the audio file if I set it up as follows?

  • initial tempo: -50%
  • final tempo: 0%
  • initial pitch shift: -12
  • final pitch shift: 0
    I think the tempo change and pitch shift are independent if I understand the online manual correctly. It uses SBSMS algorithm.
    However, the sound quality is quite good as if the sliding is resulted in as done by resampling.

I really appreciate any help you can provide.

No.
Audacity has two different time-stretch algorithms.

One is provided by the “SoundTouch” library. This implements a kind of “overlap add” called “SOLA” (Synchronous-OverLap-Add). This method is fast, but may produce noticeable and unwanted echo-like artefacts.

The other is provided by the SBSMS library. This implements a different technique called octave sub-band sinusoidal modelling. In short, this technique analyses the spectrum of the sound, then synthesizes the “stretched” sound based on that analysis. This method requires a lot of number crunching, so it is a lot slower than SOLA, but does not suffer from the echo-like artefacts.

In Audacity’s “Change Pitch” and “Change Tempo” effects, the SBSMS method is used when the “Use high quality stretching (slow)” option is enabled, otherwise it uses SOLA.
The “Sliding Stretch” effect always uses SBSMS.

Thank you very much! The SBSMS is very impressive!