Quality and BitRate of SAT Recordings posted on tribal mixes, page 2

be on it all you want, this conversation is getting boring to me... this whole thing effects me in no way, because i dont listen to XM to start with, so im not gonna be involved in this bicker anymore..

You guys work this out on your own, and if you still feel its worth 320kbps then do it to it... i really dont care...

Hi,
First, I do not mind to answer the questions you ask me. But not right know, maybe later on.
Second, while we talk different languages we will not understand each other. Let’s take “but if you are NOT recording from DIRECTtv” for an example of something I do not understand. I have said and I have clarified I record the XM Satellite Radio Channel 80 broadcast. And again, it means to me: the receiver I use receives the satellite signal, I do not now anything about it, processes it and outputs a PCM bit stream at 1,411.2 kbps on its digital output. So, what is DIRECTtv there in common?
Now I would like to point out another interesting point already mentioned by Bluemax: “low dynamic range”.
Is this phenomenon present in the XM Satellite Radio Channel 80 broadcast? I do not think it is present.
The CD rip dynamic range, at the time interval when the spectrum is calculated is about 68 dB. The XM record dynamic range is about 63 dB. If we accept these data as representative for whole signal then the difference of 6 dB is a (conditional) ratio about 1.78 which could help saving one bit to represent the digital data. Of course this helps compressing algorithms also. Based on that fact the choice of 256 kbps for an mp3 compression is not a random choice.

Hi,
To make this discussion more useful I would make some generalizations.
It is clear what "CD Quality" audio is for a record. It is 44,100 samples per second of 16-bit audio data in 2 channels
What is "CD Quality" audio for a playback which was and is still not well defined?
As I said in other words my assumption is the record has utilized the record "CD Quality" almost in full.
Then:
1) the audio spectrum bandwidth should be let’s say more than 20 kHz (the limit is 22.05 kHz  defined by the sampling rate)
2) the dynamic range should be no less than 60 dB and the minimum SQNR more than 20 dB (SQNR – Signal to Quantization Noise Ratio) which is defined by the 16 bits audio data (the limit for the sum of both is 90.3 dB).

In other words – the "CD Quality" is defined by the audio signal bandwidth and dynamic range. So, it is not subjective matter but fully and objective measurable. As to how the audio signal sounds it depends on additional factors also.

Under such a definition the mp3 compression @192 kbps can never be a CD quality, but @256 and @320 could be.
As long as, forced by criticism I faced when uploading XM channel 80 broadcasts @320 kbps, I have checked a couple of files compressed @320 kbps I had downloaded through the tracker and I have found there were files which quality was no better than of a 192 kbps compression, I would like to say one more thing that could be helpful for the uploaders.
It is obviously that sharing files @320 kbps mp3 compression is, as it was said, a "waste of bandwidth and time" provided that the sound is not a "CD Quality". (Let’s say with spectrum bandwidth of 16 kHz.)
It is better to recompress the file at a bit rate that preserve the original bandwidth. (Unfortunately it is not so easy to correct a bad dynamic range compression.)
The correct, lossless procedure should be:
First: do file format conversion to uncompressed format (wav) which is lossless and then
Second: compress the file at desired, lower bit rate.
If the original compressed file is to be compressed again, even at a bit rate that could preserve the original quality, most probably some quality would be lost. And to stress it: every recompression of a mp3 file, no matter to higher or lower bit rate, most probably will result in a quality loss.

I was asked “please tell us if u hear a difference” between 192 kbps and 320 kbps mp3 compressions.
I am not ready to conduct such tests. To answer the question I am going to use a reference, your preferable approach, this time.
Here is what Sam C. Lin says (https://www.lincomatic.com/mp3/mp3quality.html) when discussing sound quality of mp3's vs audio CD's :
“,,,I feel that 256Kbps is the best bit rate to use, because it sounds significantly better than 192Kbps.”
“I couldn't hear any improvement at all when going to up to 320Kbps” (from 256 kbps)
I agree with the second statement as long as looking at the spectra I would say: I believe I could not hear any differences also.
As to the first statement I would say there is a significant difference between the spectrums also. And the significant difference is the spectrum bandwidth. It is usually 16 kHz  - for 192 kbps and 21 kHz - for 256 kHz provided that the source was CD quality sound, the way it is defined above.

I can say here only one thing the thing that matters here is the music and mixing technique.

Quality for me means quality tracks and beats and mixing skills.

And for some mixes I'm glad to have them even if they are on 128 as long as I can listen to the music and the quality is not crapy (not the bitrate but the quality of the recording is that matters)

I mean I'm not a sond engineer and my ears are not so sensitive(so much satying in the club next to the speakers ya know  )

So for this whole topic is not necessary as long as we got the quality in music the bitrate is a secondary thing      

Yeh, I like 128 & 192...they are good quality(as sound) and they are not so big like 320 one's :)

I have posted two more spectra of compressed @320 kbps sounds. The first one is (10) Shaun King.mp3. It is from the album Circoloco@DC10 mixed by Tania Vulcano. (If you do not have it, take it: https://www.tribalmixes.com/details.php?id=14849).
The second one is John Digweed at Transitions on The Move (XM Sat) 22-Feb-2007(1).mp3.
It is not difficult to see: the first one is CD rip, the second one is XM Sat Radio record.
Formally speaking both spectra are CD quality sounds. But the drawbacks of the XM Sat Radio record are easy to find.
Let us discus them.

my 2¢...

encoding a satellite transmission or internet radio show in any bitrate above what it's broadcast in will ALWAYS:

a) reduce sound quality
b) waste hd space / bandwidth / time

there is absolutely no reason whatsoever to do so!  you can't magically improve the sound quality by increasing the bitrate-- it just doesn't work that way!

Of course, you are right. Having in mind: “bit rate above what it's broadcast” what has left is for you to tell: what is the proper bit rate to encode the XM Sat channel 80 broadcasts.
P.S. Or in other words, what is the XM Sat Radio channel 80 broadcast’s bit rate?

The main topic - “Quality of the XM Sat Radio Cannel 80 Broadcasts”. It depends on two factors: the sound generator and the communication channel. I am going to show that the  XM Sat Radio Cannel 80 communication channel can preserve a CD quality sound if it is transferred through it. The way to do this is to show at least one spectrum generated over the data recorded from a  XM Sat Radio Cannel 80 broadcast that has CD quality properties.
This time I have to use quantitative evaluations, so let me specify them:
Spectrum bandwidth: 22,050 Hz.
Spectrum dynamic range more than 60 dB.
Signal to Quantization Noise Ratio (SQNR) more than 20 dB.
The first two properties are already known. Why I need SQNR? For the simple reason to have a spectrum bandwidth in which the smallest components are considerably bigger than the quantization noise.
First, why 20 dB. To go for better simply means to use 24 bits audio insted of 16 bits audio.
Second, 20 dB means ratio of 10.
To specify SQNR I need to specify the Quantization Noise level.
Now, the more difficult part. What I am going to state is based on my knowledge. So, it can be wrong and I do not what to bear any responsibilities for your using it.
Shortly, the approach to quantization error evaluation is to consider it as a white noise. A white noise with amplitude of a maximum quantization error, which is ½, has a constant spectral density of 1/12. Then the Quantization Noise level is 20log (1/12* 32768) = -111.8926 dB. Let me work with -112 dB.

In words:
the spectrum has to reside over the level of -112 – (-20) = -92 dB for whole frequency range of 22,050 Hz and has to have a dynamic range of at least 60 dB.

The spectrum shown below is from the Jimmie Page show on XM Sat radio on 26-Feb-2007. The spectrum has been calculated at a certain moment of time using 65,536 samples, which means it is calculated over a discrete time signal of 1.486 seconds.

So,
is it a spectrum of a CD quality sound?
Yes, it is.
P.S. As to the XM bit rate an estimation I could make is: it is higher than 256 kbps and obviously less than (maybe) 0.65*1,411.2 = 917 kbps.
But this is another topic

hmm... interesting. didn't think XM was broadcasted @ such a high bitrate maybe something close to ¾ of that bitrate is possible on the DirecTV XM stations...

I need to discuss two more topics: the quality of the XM Sat Radio cannel 80 broadcasts and the bit rate of these broadcasts.
The full information about the sound signal is contained in its sonogram. Look at sonogram below. A sonogram represents the sound signal spectrum along the time. So, it is 3D information (time, frequency, and spectrum value) shown as 2D plot. The third dimension is color coded.
It is obviously it is very difficult to examine the sound quality.  More over, we are interesting in communication channel properties that influence the sound quality, rather than in sound quality itself.



That is why I’ll try to show some channel peculiarities that are relevant to the sound quality and the bit rate.
What I am showing next is a sound spectrum generated to show rather qualitative than quantitative way the channel properties. (I am saying that because it is not a real spectrum but again an “averaged” one over a time interval of 30 seconds only to make it "smooth".)



The main peculiarity is the spectrum slope of 20 dB/decade over the mid frequency range (1000 Hz-10,000 Hz), shown with a red line.
If it is true, it means XM use a filter to decrease the signal. I guest the purpose of decreasing the signal level is to “save” bits when transferring the data through the channel. In other words to decrease the bit rate they use.
There are three more peculiarities marked by ellipses. I think they are side effects of the filter implementation. Two of them, at 5.4 kHz and at 11.1 kHz are drawbacks, the third one, at 15 kHz, shows they intentionally suppress the spectrum high frequency range (15 kHz – 22 kHz). The signal level drop is around 8 dB or 2.5 times (as it can be evaluated by the spectrum generated for the signal at 00:24:40.132 shown next).



If I am right, the channel preserves the signal bandwidth but the radio suppresses the signal level at mid and high spectrum range this way preserving the sound dynamic range and creating conditions for decreasing the bit rate it uses to broadcast

Well you wrote it already,
I highly doubt that XM broadcasts at 917 kbps. This is ridiculous.
Speaking about 'WhiteNoise' i guess what u see in your 'spectra' is the standard PCM out signal you capture from. so at least would it explain how you come to 917, almost 1411.2 kbit/s
I again ask you to be so kind and capture an ordinary DVB-S MP2 satellite radio signal and make a analysis thru the very same reciever. or is only XM recieveable with this device?
Im also interested why the XM signal has a much higher value by the end of the scale at 22khz?
where does this come from?
The db scale is a log scale and a change of a few db mean a massive change in linear mode.
So a 8-10db difference is a massive difference and i dont think you cant talk about 'same quality' annymore.

btw: i have to 'obtain' the sony tool so i can try to reproduce ur graphs...

First, let me continue with the bit rate.
As to the bit rate of the XM channel 80 broadcasts.
The idea of its evaluation is: The CD quality sound is a bit stream at 1,411.2 kbps.
So, estimation from above for the XM radio broadcast bit rate is 1,411.2 kbps. (Or in other words - no need for higher bit rate than 1,411.2 kbps) Nobody believes that because using FLAC the bit rate can be reduced approximately with 65%.
Better estimation could be a bit stream generated by a compression at lower bit rate such that it has the same spectrum as its origin. Personally, I have only mp3 compressor and so, the next bit rate I can try is 320 kbps. The fact, shown above, that the spectrum of an mp3 compressed at 320 kbps sound visually does not differ from its uncompress origin over the high frequency range, can not be used to estimate from above the bit rate at 320 kbps. It is because I could be lucky to choose a “good” file. It is possible for a recorded file to exist such that it can not be compressed considerably lossless using an mp3 compression at 320 kbps. So, it is better to say, maybe realistic estimation of the bit rate from above could be 320 kbps.
Estimation from below has already been shown. It is 256 kbps. It follows by the fact that the spectrum of an mp3 compressed at 256 kbps sound differs from its uncompress origin over the high frequency range. (I saw I have not posted a spectrum of a mp3 at 256 kbps but it is not a problem, it is true.)