Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: distortion calculations (Read 2465 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

distortion calculations

I noticed that from the MP3 and AAC specs, the distortion calculation is taken as sum of "energy of error signals"..  Would it be better if it is the fabs( error of  sum energy) to better match the threshold calculated from a Psychoacoustic Model based on Model I?

distortion calculations

Reply #1 –
Quote
I noticed that from the MP3 and AAC specs, the distortion calculation is taken as sum of "energy of error signals"..  Would it be better if it is the fabs( error of  sum energy) to better match the threshold calculated from a Psychoacoustic Model based on Model I?

I don't know about model I, but both models output ISMR,

Which is a division of sum of all squared spectral components (energy)  and noise threshold (also multiplied by the spectral components)

So, the "natural" algorithm for noise calculation would be:

1. compute difference of re-quantized signal and allowed distortion
- we will call this "difference signal"

2. now measure energy of diff. signal by squaring the spectral components

3. Sum of individual error components divided by allowed dist (xmin in ISO notation) is the NMR - noise to mask ratio,  usually used as a distortion measure.


Your proposed solution is usually not good , because it is a difference, not division - but it might work if you calculate all values in dB domain (tried with AAC Q&C once, and results were not correlating with psychoacoustic - there is a test on this forum "codec_a" vs "codec_b", and "codec_b" used this error measure - I think it was in October 2002, if you search this forum you might find subjective results).