IMA vs MS ADPCM. Noise floor wrt sine wave different?

2007-04-04 14:54:27

I am trying to figure out why MS ADPCM behaves so much differently to IMA ADPCM when I do the following experiment.

1) In Audacity, create a 440Hz tone, 44.1KHz sampling rate at 80% FSD level,

2) Export in MS ADPCM wav format,

3) Export in IMA ADPCM wav format,

4) Re-import and listen with headphones.

The result is that the IMA version has a constant audible hiss in the background like a nasty cassette tape. However the MS version has no audible hiss most of the time but occationally there are busts of noise in the background at about the same level as the IMA version.

I've been looking at various source code for ADPCM implementations and the only difference I know of is that the MS implementation resets the codec every 1ms, but not at full 16bit resolution. Would that do it? Or is that a red-herrin?

Can someone repeat the experiment and let me know what they think is going on?

Thanks

IMA vs MS ADPCM. Noise floor wrt sine wave different?

Reply #1 – 2012-04-19 12:35:59

Hi bro,

          Can you explain ADPCM technique and also how it can be implemented to wav file. can u you provide any links for source code ?My application
needs this technique to be applied.I am not able get enough information on that.

Thanks in Advance

IMA vs MS ADPCM. Noise floor wrt sine wave different?

Reply #2 – 2012-04-19 14:29:50

ksr: It is often a good thing not to start new threads on topics which already has one, but don't expect answers from the original poster when the message is over five years old, and was his last post here ever