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: **The Lowest bitrate of CD quality: 32 or 16kbps?** (Read 7415 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

hi
    Does anyone think of the lowest bitrate  of CD quanlity?
   
    what about 16kbit/s/stereo?

    AAC may not success for it.
    mpeg-4 audio is possible.

    Can Ogg vorbis arrive at the aim?

    Do we need some new audio codec(that is, new algorithm)?

    who can give some comments and prodiction for next generation audio comprssion technology?


thanks.

Any fancy ideas are welcome!!!!

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #1 –
Quote
Originally posted by jianxin yan

    Does anyone think of the lowest bitrate  of CD quanlity? 


Is that CD quality as in:

1) Xing @ 128kbps
2) WMA @ 160kbps
3) MPC -standard
4) lossless

?

--
GCP

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #2 –
hi

i think CD quality should be 128kbps ogg enc now!

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #3 –
Quote
Originally posted by jianxin yan
hi

i think CD quality should be 128kbps ogg enc now!
I don´t think so... Not even -q 10 can deliver CD quality since that would require OGG to be a lossless format...

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #4 –
Quote
Is that CD quality as in:
1) Xing @ 128kbps
2) WMA @ 160kbps
3) MPC -standard
4) lossless?

Probably the same CD quality that Microsoft have with WMA @ 64kpbs. All I can say is that they must have a very poor quality CD player

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #5 –
Quote
Originally posted by Sachankara
I don´t think so... Not even -q 10 can deliver CD quality since that would require OGG to be a lossless format...


But OTOH 256kbps CBR is CD quality because the Heise test proved it to be.

Your complaint is unfounded. I asked the man what kind of CD quality he had in mind, since it's a debatable issue.

It's not because you view it differently that he's wrong. That's the key of subjectivity, and 'CD quality' is a very subjective issue.

--
GCP

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #6 –
Quote
originally posted by Garf:
But OTOH 256kbps CBR is CD quality because the Heise test proved it to be.

iirc this was some years ago, with listeners untrained to spot the flaws of lossy/psychoacoustic compression... I wonder to which extent this test might still hold true. ???
A riddle is a short sword attached to the next 2000 years.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #7 –
Quote
Originally posted by ssamadhi97

iirc this was some years ago, with listeners untrained to spot the flaws of lossy/psychoacoustic compression... I wonder to which extent this test might still hold true. ???


Joe Average is still untrained to lossy/psychoacoustic artifacts.  Do the test again, you'll get the same result.

--
GCP

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #8 –
Quote
Originally posted by ssamadhi97
iirc this was some years ago, with listeners untrained to spot the flaws of lossy/psychoacoustic compression... I wonder to which extent this test might still hold true. ???
What Garf said, with the addition that I'd defend using untrained listeners.  I want a codec tuned so that it sounds transparent to an average listener (one listening carefully, not just with music on in the background, but not one specifically trained either).  As an average untrained listener, that's the sort of quality I'd be interested in - if I can't notice a difference, then it's transparent as far as I'm concerned.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #9 –
hi

i think mpc 128 , aac 128 and ogg 128 can be accepted by most listeners.

also i donot want to argue what is CD quality, the topic seems not to be answered correctly as far.

therefore. i mainly hope to discuss what should be considered for next generation codec, and can we arrive at 16kbps/stereo with aac 128 audio quality? how many years? how to do?


thanks.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #10 –
Quote
and can we arrive at 16kbps/stereo with aac 128 audio quality? how many years? how to do?

What you find with lossless compression is that all the big reductions come relatively easily - and you spend the next 10 years chopping off 0.5% here, 0.5% there. I imagine the same is true of lossy compression.

For a general purpose codec, I'd be astonished if they had the same sound quality at 16kpbs that we current have at 128kpbs today. That's an 8-fold difference in file size -- and if it is achievable at all, it'll only come with a completely seperate approach to the ones that are currently being used. I even have my doubts about 64kpbs.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #11 –
Quote
Originally posted by jianxin yan
therefore. i mainly hope to discuss what should be considered for next generation codec, and can we arrive at 16kbps/stereo with aac 128 audio quality? how many years? how to do?
thanks.



At this point, there is an ongoing work inside the MPEG commitee towards so-called MPEG-4 V3 - this extension to the standard will have so-called "bandwidth extension" tool.

Bandwidth extension is very similar to the SBR used in MP3Pro and AAC+, but with few important differences:

1. It will be part of the ISO/MPEG-4 standard which means "open"

2. It will have best of SBR and some other technologies, which means that it will probably be better than SBR

MPEG commitee has set requirement that MPEG-4 V3 should sound like MPEG-4 at 30% higher bit rate (i.e. MPEG-4 V3 AAC at 24 kbps will sound like MPEG-4 AAC at 32 Kbs... We will see if this could get even better.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #12 –
hi

i think "bandwidth extension" is not a good method for improving audio quality at very low bitrate such as 16kbps/stereo, which is adapted to DRM with FM audio quality.

With SBR, PlusV or mpeg-4 noise substitution, 32kbps/stereo qulity maybe not arrive at current 128kbps aac.

So hope to give some creative technology for the aim of the 16kbps/stereo codec.

regards.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #13 –
I'm affraid that there is no way to achieve 128 kbps quality with 16 kbps - MPEG-4 with SBR will probably be lowest bitrate that could give at least decent (torelable?) quality...

The problems you are facing are purely mathematic, not engineering in their nature - i.e. there is no way to represent data, such as music with entropy that could represented with formulae (like perceptual entropy formula)

Compare that with elementary physics, i.e. you can't stick two objects at the same place

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #14 –
Quote
Originally posted by Ivan Dimkovic
I'm affraid that there is no way to achieve 128 kbps quality with 16 kbps - MPEG-4 with SBR will probably be lowest bitrate that could give at least decent (torelable?) quality...


yes, i also admit this viewpoint  at present.

Quote
The problems you are facing are purely mathematic, not engineering in their nature - i.e. there is no way to represent data, such as music with entropy that could represented with formulae (like perceptual entropy formula) 

Compare that with elementary physics, i.e. you can't stick two objects at the same place


Indeed, this may involve some more theoretic problem, but i still think we should strive for the aim in following 5 years. therefore we should bring forward  more new ideas.
  On this board, most of people seem to like talking about some encoder switchs, why not we talk and argue some  deeper technologies or possible  fancy ideas?

regards.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #15 –
I'll try to explain somethin'


1. Imagine a perfect model of human hearing that gives EXACT results as average human hear

2. Imagine a perfect filterbank, with perfect reconstruction (and diagonalization)

3. Imagine a perfect bit-allocation (quantization) system, with near-zero side information, which introduces exact amount of noise as allowed by the psychoacoustic model

We can define 2. and 3. - and simulate 1. , now with this simplification, bitrate required for the transparent coding will be around 1.4 bit/sample for the transparent audio coding.

Now, this is the absolute limit - and this limit can't be ultimately reached because no coding system could be ever this perfect!!!

Below ~1.4 bits (so-called 'perceptual entropy'), we can do some tricks like

- Fake stereo / Intensity Stereo...
- Spectral band replication / reconstruction

But all of these techniques are trying to simulate perfect reconstruction, and not to provide transparent quality - this is why something like '128 kb/s like' quality is impossible at bit rates of 16 kbps - maybe some tricks like in QDESIGN Music codec could help, but this codec fails in many categories.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #16 –
Quote
Originally posted by Ivan Dimkovic
I'll try to explain somethin'
1. Imagine a perfect model of human hearing that gives EXACT results as average human hear
2. Imagine a perfect filterbank, with perfect reconstruction (and diagonalization)
3. Imagine a perfect bit-allocation (quantization) system, with near-zero side information, which introduces exact amount of noise as allowed by the psychoacoustic model
We can define 2. and 3. - and simulate 1. , now with this simplification, bitrate required for the transparent coding will be around 1.4 bit/sample for the transparent audio coding.
Now, this is the absolute limit - and this limit can't be ultimately reached because no coding system could be ever this perfect!!!
Below ~1.4 bits (so-called 'perceptual entropy'), we can do some tricks like
- Fake stereo / Intensity Stereo...
- Spectral band replication / reconstruction
But all of these techniques are trying to simulate perfect reconstruction, and not to provide transparent quality - this is why something like '128 kb/s like' quality is impossible at bit rates of 16 kbps - maybe some tricks like in QDESIGN Music codec could help, but this codec fails in many categories.


What I have in mind are Wavelet transform (I read this from somewhere saying it is good for low-bitrate encoding) and Fractal Compression for Audio (I wonder whether it's possible? Researches on Fractal Compression for Graphics are still under development right?).

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #17 –
hi

  if we can't break the PE limit, we donot certainly get 16bps/stereo anyway.

I think some  tricks for  reaching 128kbps aac quality is quite difficult, expecialy for various audio clips.

Do you think PE have discribed audio information completely and correctly?
Can't we give more correct discription?


thanks.

**The Lowest bitrate of CD quality: 32 or 16kbps?**

Reply #18 –
PE is the theoretical limit - note that PE does not define how the level of masking is computed.

This is similar to the compression entropy value, defined for loseless codec - we can't zip 1 MB file to 1 byte. Codecs can only be more or less close to that value (for transparent coding), for example AAC is closer to the PE limit than MP3 is.

Of course, many other design parameters influence the quality of one codec, like temporal/frequency resolution, NMR disturbance in frequency and time, and of course the quality of the analysis (psychoacoustic) model.

Like I said, if we go below "transparency" defined by PE, it is possible to employ various algorithms for approximation, like SBR/Intensity Stereo/PNS, etc..