I found this station at live365
http://www.live365.com/stations/compaxx
The sound is really good, but their stream is only 56k. They claim their sound is better than CD.
Anyone know about their "Compaxx" technology?
What does everyone think about thier sound?
Full Version: high quality 56k stream?
QUOTE(Ishtov @ Dec 6 2006, 12:06) 
I found this station at live365
http://www.live365.com/stations/compaxx
The sound is really good, but their stream is only 56k. They claim their sound is better than CD.
Anyone know about their "Compaxx" technology?
What does everyone think about thier sound?
They're just using mp3/mp3PRO ? That's what it looks like to me anyway
Just listen to some of the 'other' channels like the rock stuff and it's sounds blooming terrible at the same 56Kbps..
From what I understand, it's regular mp3 (otherwise you'd need an mp3pro-capable decoder on the software side) but they pre-process the audio before encoding to reduce the typical mp3 artefacts. This thread touched on a similar idea.
The big question is whether those test samples are comparable to all music, or just extremely favourable examples.
The big question is whether those test samples are comparable to all music, or just extremely favourable examples.
It's amazing what you can achieve with HE-AAC in streams nowadays. Listen to this 48kbps stream for instance.
However, I checked the 365 stream and it is not an AAC stream of any sort. Indeed it looks more like a (modified) MP3 stream.
However, I checked the 365 stream and it is not an AAC stream of any sort. Indeed it looks more like a (modified) MP3 stream.
QUOTE(Ishtov @ Dec 6 2006, 13:06)
I found this station at live365
http://www.live365.com/stations/compaxx
The sound is really good, but their stream is only 56k. They claim their sound is better than CD.
Anyone know about their "Compaxx" technology?
What does everyone think about their sound?
I came across this post while crawling, and would like to thank Ishtov and others for your interest in Compaxx mp3. I'm a bit short on reply time right now, but would be delighted to pursue this thread if it's of general interest. To anwer a basic question raised here, no, Compaxx does NOT utilize mp3PRO, or require any external plugins.
If you'd like to hear comparison samples of Compaxx vs. original CD tracks, please visit http://www.compaxx.org. This is a broad selection of genres and quality levels, and give a good cross-section of Compaxx performance.
Tuvia Wertheim,
Compaxx DHF Masterworks
This should be fun . . .
Just checked the "Nothing else matters" comparison sample from the website.
I mean - come on, this is supposed to be BETTER?
The bass is bloated and the trebles are hissing (bathtub EQ), and apparently there has been a whole lot of dynamic compression.
It is stated there that all samples are peak normalized, however, these compressed versions carry a much higher signal energy than the CD ones. From ABX methodology it is known that contenders must be level matched. Comparing signals with so apparently different perceived volume is pretty pointless since it is a dead giveaway. Level differences (for same source material, mind you) should be smaller than 0.1 dB. This is because often louder is being perceived as better.
(Look at the dynamics statistics in a wave editor, or even at the waveform itself, it is so apparent).
As for the EQ, this *might* cause the music to sound better on cheapo (computer, laptop, OEM car) speakers, but then again, this is why the original on the CD is mastered so vanilla: everybody can, starting from the most neutral version, accomodate for their own sound system.
As one might have guessed by now, I prefer the CD version
I would check the stream mentioned in the first post, but this site seems to require a proprietary, windows-only client... no way
In order to evaluate how 'good' the processed version of the streamed audio really is, one should of course have access to a non-processed, same encoded version of the audio.
I mean - come on, this is supposed to be BETTER?
The bass is bloated and the trebles are hissing (bathtub EQ), and apparently there has been a whole lot of dynamic compression.
It is stated there that all samples are peak normalized, however, these compressed versions carry a much higher signal energy than the CD ones. From ABX methodology it is known that contenders must be level matched. Comparing signals with so apparently different perceived volume is pretty pointless since it is a dead giveaway. Level differences (for same source material, mind you) should be smaller than 0.1 dB. This is because often louder is being perceived as better.
(Look at the dynamics statistics in a wave editor, or even at the waveform itself, it is so apparent).
As for the EQ, this *might* cause the music to sound better on cheapo (computer, laptop, OEM car) speakers, but then again, this is why the original on the CD is mastered so vanilla: everybody can, starting from the most neutral version, accomodate for their own sound system.
As one might have guessed by now, I prefer the CD version
I would check the stream mentioned in the first post, but this site seems to require a proprietary, windows-only client... no way
In order to evaluate how 'good' the processed version of the streamed audio really is, one should of course have access to a non-processed, same encoded version of the audio.
lol, the so-called CD-quality samples on their site are encoded in mp3 (192 kbps CBR LR stereo FhG)
And of course, the qualification 'better than CD-quality' is meaningless. Any attempt to go beyond transparancy implies changing the original work of art.
And of course, the qualification 'better than CD-quality' is meaningless. Any attempt to go beyond transparancy implies changing the original work of art.
Compaxx is not at all about the quality (i.e. transparency) of the mp3 encoded files.
As ImAlive has already reported, Compaxx is just an audio enhancer. The "secret formula" (i.e. DSP algorithms) of Compaxx will make it sound better than the original CD recording.
Btw. my favourite search engine pointed me to a discussion on stereophile.com about this topic. I found this especially interesting:
Magic mp3's!
As ImAlive has already reported, Compaxx is just an audio enhancer. The "secret formula" (i.e. DSP algorithms) of Compaxx will make it sound better than the original CD recording.
Btw. my favourite search engine pointed me to a discussion on stereophile.com about this topic. I found this especially interesting:
QUOTE
Compaxx (Tuvia) at forum.stereophile.com
> Decoding to WAV for analysis: You're welcome to try this, but bear in mind that a significant amount of the embedded subcode risks damage in the process (in particular the spatial cuing and subwoofer subtracks). For best results I'd recommend analog-playback analysis, or digital i/o reads taken directly off the decoder chipset.
> Decoding to WAV for analysis: You're welcome to try this, but bear in mind that a significant amount of the embedded subcode risks damage in the process (in particular the spatial cuing and subwoofer subtracks). For best results I'd recommend analog-playback analysis, or digital i/o reads taken directly off the decoder chipset.
QUOTE
Compaxx (Tuvia) at forum.stereophile.com
a) Peak Volume Cutoffs: I'm not surprised that, after decoding to wav despite my warning, you produced a few (4) minor peak cutoffs in the Vivaldi sample. When stripping the mp3 envelope during decoding you've removed critical cuing information, resulting in a 2.2% increase in volume level with resulting cutoffs. Therefore, when I produce Compaxx in PCM format for CD application, I do so at a 2.2% lower peak volume than when producing Compaxx mp3. (On the Compaxx Benchmark CDROM I present both versions. When measured WITHIN the mp3 envelope, the peak volume is idential to the CD version.)
a) Peak Volume Cutoffs: I'm not surprised that, after decoding to wav despite my warning, you produced a few (4) minor peak cutoffs in the Vivaldi sample. When stripping the mp3 envelope during decoding you've removed critical cuing information, resulting in a 2.2% increase in volume level with resulting cutoffs. Therefore, when I produce Compaxx in PCM format for CD application, I do so at a 2.2% lower peak volume than when producing Compaxx mp3. (On the Compaxx Benchmark CDROM I present both versions. When measured WITHIN the mp3 envelope, the peak volume is idential to the CD version.)
Magic mp3's!
QUOTE
Compaxx (Tuvia) at forum.stereophile.com
> Decoding to WAV for analysis: You're welcome to try this, but bear in mind that a significant amount of the embedded subcode risks damage in the process (in particular the spatial cuing and subwoofer subtracks). For best results I'd recommend analog-playback analysis, or digital i/o reads taken directly off the decoder chipset.
This is just so great. Data being decoded from mp3 to PCM and written to a file is different from data being decoded from mp3 to PCM (through the same ISO decoder) and written to an output buffer for playback. And what is cuing?> Decoding to WAV for analysis: You're welcome to try this, but bear in mind that a significant amount of the embedded subcode risks damage in the process (in particular the spatial cuing and subwoofer subtracks). For best results I'd recommend analog-playback analysis, or digital i/o reads taken directly off the decoder chipset.
For the second quote: Peaks in mp3s can indeed be higher than in the original, especially with flat-top limited music, but I don't quite understand what the point of the quote should be...
QUOTE(Compaxx DHF Producer @ Dec 25 2006, 19:35)
If you'd like to hear comparison samples of Compaxx vs. original CD tracks, please visit http://www.compaxx.org.
BTW this is a HA.org TOS #9 violation. Maximum sample length for copyrighted music is 30 seconds.
...i'm fairly sure they have permission
It looks like just dynamic range compression, a volume increase, and EQ. I'd rather not have dynamic range compression (most CDs already have too much) and apply my own EQ that works well on my speakers.
Thank you all for your above comments. Please note that I am currently updating the Benchmarks page posted above.
In the meantime, please feel free to use the following link to stream 'Before & After' samples of Compaxx DHF mp3: http://www.compaxx.org/php/benchmarks.php. Click on the green icons to play any individual or group of tracks. The classical section also includes a few full-length tracks for extended listening.
Please note that all samples are in 192k mp3 format. The files noted as 'CD' are compressed directly from CD-wav rips, with no intermediate process. The reason I've posted these as opposed to the original wav files is that PCM is chunky and not readily streamable.
A few brief notes on thread comments above:
0. Ishtov - Thanks for kicking off this discussion. Hope you enjoy our show on Live365. We'll be expanding our own classical network shortly in terms of libraries, features, and bandwidths, so stay tuned!
1. m00 - No, Compaxx is not mp3PRO. Our 56k version was specifically designed to obviate the neccesity for external plugins such as mp3PRO.
2. ImAlive - re: Metallica: - The Compaxx process does NOT use any volume or EQ compression. The process is based on multi-level restoration via phasing and data-mining of dormant audio information resident within PCM. If suppressed background gunk reappears during restoration, so be it.
RMS (amplitude) issues - As for apparent RMS differences, were I to raise the levels within the original CD tracks, they'd no longer be original, as this is already a significant change. Also, without significant volume compression (a no-no), it is impossible to raise the RMS of the originals to Compaxx levels without severe clipping. Try this yourself if you wish.
98%-peak w/2% gain while decoding: Software mp3 decoders boost peak levels by about 2%, hardware chipset decoders generally do not. Therefore, if you wish to analyse Compaxx files via software, please peak-normalize to 98% after decoding to wav before analysis. This Stereophile comment was a response to John Atkins, who had noted a few minor peak cutoffs in a Vivaldi sample while software-decoding.
3. Hanky - Regarding 'original art', I don't consider CD renderings to be original art; the original live performance is my standard. All of my pre-compression rendering routines are based on actual live performance models (analog), not CD. I've spent over 35 years as a musician, producer, and broadcaster, and am seeking here to recover the warmth, depth, and spatial presence of the live performance, while retaining the detail of digital. Wherever possible, I rely on direct feedback from the performing artists and conductors as to the authenticity of the sound achieved.
4. smack - Thanks for the positive input. Regarding the Stereophile posts, they're a bit old and I'll have to revisit and refresh there. The comments were based at the time on a prior version, and I've redesigned based on the feedback from there. Stereophile was tremendously helpful.
5. ...Just Elliott - Sample length. It is nearly impossible within a 30-sec. sample to find a section of a recording which is representative of the work as a whole. That aside, as a veteran industry professional I am exceedingly sensitive about royalties and DMCA issues. In addition to paying royalties where appropriate, I've submitted all of our samples to the rights holders (record labels) for review. Should any rights holder not be happy with our posting of samples, I'm willing to either review our royalty agreements or alternatively, remove said samples.
6. slks - Again, we don't use volume or EQ compression, or any type of reverb. Please analyse and note the total absence in Compaxx of bandwidth markings or top-heavy peaks, which would be the obvious result of said techniques when applied.
That does it for now. Again, thank you all for your comments. MP3 is here to stay; my goal is to make it perform better, especially in broadcasting for classical listeners.
Best to all in the New Year,
Tuvia Wertheim,
Compaxx DHF Masterworks
In the meantime, please feel free to use the following link to stream 'Before & After' samples of Compaxx DHF mp3: http://www.compaxx.org/php/benchmarks.php. Click on the green icons to play any individual or group of tracks. The classical section also includes a few full-length tracks for extended listening.
Please note that all samples are in 192k mp3 format. The files noted as 'CD' are compressed directly from CD-wav rips, with no intermediate process. The reason I've posted these as opposed to the original wav files is that PCM is chunky and not readily streamable.
A few brief notes on thread comments above:
0. Ishtov - Thanks for kicking off this discussion. Hope you enjoy our show on Live365. We'll be expanding our own classical network shortly in terms of libraries, features, and bandwidths, so stay tuned!
1. m00 - No, Compaxx is not mp3PRO. Our 56k version was specifically designed to obviate the neccesity for external plugins such as mp3PRO.
2. ImAlive - re: Metallica: - The Compaxx process does NOT use any volume or EQ compression. The process is based on multi-level restoration via phasing and data-mining of dormant audio information resident within PCM. If suppressed background gunk reappears during restoration, so be it.
RMS (amplitude) issues - As for apparent RMS differences, were I to raise the levels within the original CD tracks, they'd no longer be original, as this is already a significant change. Also, without significant volume compression (a no-no), it is impossible to raise the RMS of the originals to Compaxx levels without severe clipping. Try this yourself if you wish.
98%-peak w/2% gain while decoding: Software mp3 decoders boost peak levels by about 2%, hardware chipset decoders generally do not. Therefore, if you wish to analyse Compaxx files via software, please peak-normalize to 98% after decoding to wav before analysis. This Stereophile comment was a response to John Atkins, who had noted a few minor peak cutoffs in a Vivaldi sample while software-decoding.
3. Hanky - Regarding 'original art', I don't consider CD renderings to be original art; the original live performance is my standard. All of my pre-compression rendering routines are based on actual live performance models (analog), not CD. I've spent over 35 years as a musician, producer, and broadcaster, and am seeking here to recover the warmth, depth, and spatial presence of the live performance, while retaining the detail of digital. Wherever possible, I rely on direct feedback from the performing artists and conductors as to the authenticity of the sound achieved.
4. smack - Thanks for the positive input. Regarding the Stereophile posts, they're a bit old and I'll have to revisit and refresh there. The comments were based at the time on a prior version, and I've redesigned based on the feedback from there. Stereophile was tremendously helpful.
5. ...Just Elliott - Sample length. It is nearly impossible within a 30-sec. sample to find a section of a recording which is representative of the work as a whole. That aside, as a veteran industry professional I am exceedingly sensitive about royalties and DMCA issues. In addition to paying royalties where appropriate, I've submitted all of our samples to the rights holders (record labels) for review. Should any rights holder not be happy with our posting of samples, I'm willing to either review our royalty agreements or alternatively, remove said samples.
6. slks - Again, we don't use volume or EQ compression, or any type of reverb. Please analyse and note the total absence in Compaxx of bandwidth markings or top-heavy peaks, which would be the obvious result of said techniques when applied.
That does it for now. Again, thank you all for your comments. MP3 is here to stay; my goal is to make it perform better, especially in broadcasting for classical listeners.
Best to all in the New Year,
Tuvia Wertheim,
Compaxx DHF Masterworks
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.