Time Stretch Trick

Hi guys,

Just wanting to share a trick I’ve just accidentally discovered while attempting to convert 192 kHz 32bit into 44.1 kHz 16bit using Cubase.

When importing the audio I forgot to change the project sample rate and bit rates, from my default 48 kHz 32bit. I then added an eq to the master channel to shelve off the bottom and top at 20hz and 20kHz respectively. Next the Apogee Dither set to 16bit Hi. selected the range and exported the mix-down at 44.1kHz 16Bit. I ended up with a beautifully smooth time stretched (45mins into 4 hours) version of the audio. As the Audio was interpreted as 48kHz and not 192 Khz.

This trick might be old news to some of you, but wanted to share the technique for those still learning, or that had not discovered it yet.

LOL! Yeah, that’s what happens when you take something at one sample rate, and try to play it at another!
I know this effect all to well from my attempts at coding an audio system for a game…

It was heavily pitch shifted too, right? I would assume so, since that’s always what happens when you re-interpret the sample rate with no pitch correction. Without sample rate conversion (either blending or removing samples) you need some sort of time-stretch logic to get the result to sound the same pitch.

Try doing that to a clip of a fork on a wine glass, or a squeaky door hinge! When slowed down, those sound very weird.

Yes was pitched down as well, sounds like playing a tape back at 1/5 its speed! It has all sort of noises chairs been moved around doors closing etc mixed in with organ music and singing / speech. I actually fact it was reel to reel tape capture of my parents wedding, from about 35 years ago!

You can achieve a similar smooth result using the built-in time stretch algorithms that have “Tape” in their name.

If you fancy some very smooth, low-artefact, extreme time stretching without affecting pitch, try this: http://hypermammut.sourceforge.net/paulstretch/