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: How to measure SNR of vorbis n mp3 (Read 7600 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

How to measure SNR of vorbis n mp3

Is there any software or tools used to measure SNR (Signal to noise ratio) of vorbis and mp3?
Is Adobe Audition can do that?
If anyone know or have the software(to measure SNR) please share it with me. I will very thanks for that.

How to measure SNR of vorbis n mp3

Reply #1 –
Why do you want to do this?

I would expect the SNR to be very dependant on the actual sample, and in the range of 10-13dB for typical pop/rock material.

How to measure SNR of vorbis n mp3

Reply #2 –
Quote from: Garf on
Why do you want to do this?

I would expect the SNR to be very dependant on the actual sample, and in the range of 10-13dB for typical pop/rock material.


I want to measure both of it for my final paper in my college. I want to make comparison of mp3 and vorbis (except listening test).Do you know anything related with that?

How to measure SNR of vorbis n mp3

Reply #3 –
Wouldn't something like this work?
1. Load original sample into some audio editor
2. Load encoded sample (make sure encoder's offset is compensated)
3. Invert phase of one of them
4. Mix
5. Measure level of resulting sample

How to measure SNR of vorbis n mp3

Reply #4 –
Quote from: cahklowor on
Quote from: Garf on

Why do you want to do this?

I would expect the SNR to be very dependant on the actual sample, and in the range of 10-13dB for typical pop/rock material.


I want to measure both of it for my final paper in my college. I want to make comparison of mp3 and vorbis (except listening test).Do you know anything related with that?


I think Garf's point is that SNR is meaningless for perceptual audio encoders, and so its not interesting to calculate.

A codec with high SNR would be very, very easy.  Just lowpass the audio signal until you hit your target bitrate.  Thats going to give you the highest SNR possible for a given bitrate (or at least the highest I can think of), but also probably the worst audio quality

How to measure SNR of vorbis n mp3

Reply #5 –
I agree with Garf. It cannot be some fixed value that you can use for comparison like say, when comparing audio electronic components. In that case, you can measure the SNR of two CD players or two amplifiers and use it as a means of comparison. However, in the case of codecs, it also depends on settings like bitrate, filtering, compression level, etc. that are being used during compression.

Even if you keep other variables constant, difference in target bitrates alone used for compression will result in different SNR values.

edit: typo
Reason is immortal, all else mortal
- Pythagoras

How to measure SNR of vorbis n mp3

Reply #6 –
Quote from: Mike Giacomelli on
A codec with high SNR would be very, very easy.  Just lowpass the audio signal until you hit your target bitrate.  Thats going to give you the highest SNR possible for a given bitrate (or at least the highest I can think of), but also probably the worst audio quality

The approach you proposed would probably work well on real music samples (with comparatively small amounts of energy in the high frequencies), but wouldn't, in my opinion, be optimal. As far as I can see, for general signals, the best way to preserve SNR while reducing bitrate would be to reduce sample size. For example, converting to four bit samples would keep your SNR at ~24dB - which is clearly better than MP3 or Vorbis (or even HE-AAC). Four bit Linear PCM is the next generation of cutting edge audio codecs

The point is, as everybody has said, SNR is a very poor performance metric for perceptual compression schemes (for sound or images). For something like MP3, as Garf said, it's likely to be awful. MPEG4 has an even worse effect on the SNR of a video stream.
 

How to measure SNR of vorbis n mp3

Reply #7 –
Quote from: cabbagerat on
The approach you proposed would probably work well on real music samples (with comparatively small amounts of energy in the high frequencies), but wouldn't, in my opinion, be optimal. As far as I can see, for general signals, the best way to preserve SNR while reducing bitrate would be to reduce sample size. For example, converting to four bit samples would keep your SNR at ~24dB - which is clearly better than MP3 or Vorbis (or even HE-AAC). Four bit Linear PCM is the next generation of cutting edge audio codecs


Nice correction, but still wrong 

The problem is that the codecs have a quite advanced way to remove redundancy (DCT + entropy coding), which won't be present in the simple Linear PCM scheme. To get an optimal SNR for a given bitrate it is more likely to be better to use the techniques as they are used by lossy hybrid codecs, such as WavPack, i.e. use a form of prediction + advanced entropy coding (MPEG 4 SLS, using IntMDCT + arithmetic coding is probably even better). Be sure to turn off any kind of noise shaping or anything remotely looking like a psymodel, since that would "degrade" the result.