Explanation of Hitpoint detection algorithm

Hi all

I’m currently doing some research that requires using hitpoints to separate a lot of audio files (e.g. for a machine learning dataset), and i was wondering if there was anyone who could give me an intuitive understanding on how the hitpoints are calculated.

Specifically, using the default settings, are the hit points located at the first zero crossing point before the intensity threshold has been reached? Also, is there a value associated for the default intensity setting (e.g. -60dB)?

Any help is greatly appreciated!

I would think you’re effectively asking Steinberg to hand over their IP.

Not really…I’m not after their code, just a broad example of what the hit points represent. The example I’ve given above is a widely accepted approach to event detection in DSP, which pre-dates the function in Cubase.

I could just as easily code my example in Python - that wouldn’t mean I own the copyright to the approach, only the copyright in my computer code.


I believe, the algorithm is even much more sophisticated and it takes even the “attack time” in account. At least this is a parameter I would also take in account, if I would design this kind of feature.

Look for “onset detection” using methods such as sudden changes in spectral content. One academic source is “Music Data Analysis”, CRC Press, ISBN: 978-1-4987-1956-8

1 Like