Quality and BitRate of SAT Recordings posted on tribal mixes, page 3

Let me answer the previous questions.
1) I think it is now understandable where this value 917 came from. It is 65% (which FLAC has) from 1,411, a not precise but possible estimation of  XM bit rate from above.
The more realistic but not strict estimation for the XM bit rate is 320 kbps.
In other words, I think the XM bit rate equivalent is between 256 kbps and 917 kbps and again, more realistic, around 320 kbps.
2) I simply can not capture DVB-S MP2 satellite radio signal. I do not have a receiver. The receiver I have is an XM Radio receiver only. (I have provided information about it.)
3) The Sony Sound Forge 8.0 can be used for trail purpose free for 30 days. So, go and download it.
4) The spectrum value at the end of this particular spectrum is not high. It is -88 dB which is 1.3. The minimum possible absolute value for the 16 bits audio is 1.
5) I have shown a 6-8 dB (2 – 2.5 times) drop at 15.3 kHz which is typical for all spectra. I have said it was additional decrease of the signal level over the high frequency range and it was intentional suppress of the spectrum. It is not a normal spectrum drop. My interpretation was “bit saving” or in other words they might use some kind of encoding based on signal level over the frequency range. Actually this signal drop is the most difficult problem for me. I still think  the signal level after this drop is still high, relatively the quantization noise, so as for the signal components in the range 15 kHz – 22 kHz to be consider present in the spectrum.
6) Sorry to say that but when “Speaking about 'WhiteNoise'” you do not have to “guess”. You have to know that the “standard PCM out signal” is a digital signal representing an analog signal and this representation has a quantization error and for its evaluation the quantization error is modeled as an additive discrete time “White Noise” signal.
7) Saying “So, it can be wrong and I do not what to bear any responsibilities for your using it” is a standard way of saying “You are solely responsible for your own actions”.  I have shown my calculations I have not borrowed them. They are mine. I have said I could make mistakes. I think my calculations are right. I am not going to write a scientific paper about this topic, so I am not going to work over checking them precisely.

Maybe, the last thing I have to say.
When I offered “flac” format files I thought this was the only format entirely preserving the sound quality. Thanks to the criticism I had to check the recorded file spectra and I could see the spectra peculiarities and I understood – yes, there was no need, in order to preserve the XM Radio channel 80 broadcast sound quality, to have used “flac” format files. The mp3 compression at 320 kbps was enough
FLAC is a lossless compression though, mp3 is not.

ya know..... this argument has really gone on way too long.... if u think the quality is good then so be it... i just think personally that ppl would rather deal with a filesize of a 192bitrate set, than a 320, because there is a difference dont get me wrong, but i really dont know if its ENOUGH to download twice the size file... but thats just me... i personally dont mind 128.. hell im used to it anyways... i have good ears, but i dont sit and listen for every speck of noise or anything.. im too busy ENJOYING the music!    

Just keep doing what ur doing bro... this whole thing was never intentionally started to cause an argument over it. :)

but--- the only problem with this is that they only call it "CD quality" when the bitrate is actually more like 64kbps (i could be wrong, but it's most definitely not more than 96kbps) on regular XM, & probably about 192kbps on DirecTV's XM feeds.  so, still, there's absolutely no reason to go above these particular bitrates when recording shows from XM.

I am showing another two spectra. They have been calculated over the data at almost the same moment of the time, as precisely as I could determine it, of the last John Digweed show (received by: Kiss100 25-Feb-2007 – 192 kbps and XM ch80 01-Mar-2007 – 320 kbps).
As I have already said I consider a spectrum signal component considerable if it is greater than -92 dB (20 dB over the quantization error level of -112 dB). The level of -92 dB is shown as a red line. So, what is seen below the level -92 dB in the Kiss100 25-Feb-2007 spectrum should be considered a too small noise which is result of calculations.
The spectra are essentially the same over the frequency range of about 16 kHz. The main difference is, as it can be seen easily, the frequency range of 15.7 kHz – 22.05 kHz where the signal components in the 192 kbps compression are cut off – they are missing.
Both compressions show (almost) the same signal dynamic range of 66 dB, the 192 kbps compressed signal of the Kiss100 record has a spectrum bandwidth of 15.7 kHz (cut off frequency at -92 dB) while the 320 kbps compressed signal of the XM Radio channel80 record preserves the full possible signal bandwidth of 22.05 kHz. The file sizes should be in a ratio 320/192=1.666 or 66.7% increase in the file size which actually is the price to be paid for the last 6 kHz signal bandwidth.
P.S. The Kiss 100 radio is FM radio which tells me its bandwidth is restricted to 16 kHz. It is obvious that the XM Sat Radio Cannel80 is not restricted to less than 22.05 kHz.
As to the its equivalent bit rate, the last 320 kbps spectra is a little bit worse than this of a corresponding uncompressed file I have recorded. I could conlude  the equivalent bit rate is higher than 320 kbps but not to much.

CD quality sound - not again
This is a reference taken from https://www.tiscali.co.uk/reference/dictionaries/computers/data/m0045785.html
"CD-quality sound - Digitized sound at 44.1 KHz and 16 bits, the standard defined in ISO 10149, known as the Red Book." It should be added: "The sound is two channel" and the standard is based on Sony/Philips documents named "The Red Book". A bit stream correspondig to it is a 1,411.2 kbps bit stream.

""CD quality" when the bitrate is actually more like 64kbps"  goes after Microsoft have announced one of their sound formats which corresponds to a 64 kbps mp3 compression. So, it is a commercial trick and has nothing in common with the real CD quality format.

"but--- the only problem with this is that they just call it "CD quality" when the bitrate is actually more like 64kbps (i could be wrong, but it's most definitely not more than 96kbps) on regular XM, & probably about 192kbps on DirecTV's XM feeds.  so, still, there's absolutely no reason to go above these particular bitrates when recording shows from XM."

If someone can say that, after having read the discussion, the only thing I can say is:
I am SORRY. No more.

the cutoff is also visible at the xm pictures at around 15khz no doubt.
so the only difference to the mp3 with the cutoff is only the mp3 encoder did a better job!
by taking the PCM signal from ur reciever there is no clear cutoff anymore.
remember Karlheinz Brandenburg?!


i guess we are walking in circles here ...
since the fact the fukin XM signal is encrypted .... hey wait this is it!
damn im so blind!
is it possible to capture the raw and encrypted sat signal from XM with an ordinary DVB-S device in a PC ?!?!?!?!
if yes: record the encrypted stream, stop the time, measure the datasize and et voila: this is ur bitrate ...

all ur nice graphs are not compareable with a tru digital raw captured data!
thats due to the fact it comes from a PCM format source.

hey spasv
i recieved an, of corse, unofficial bitrate value of xm radio.
these alters from 64kbit/s to 96kbit/s HEC aacPlus v1

again i have to laugth at xm for being so childish not officially naming the bitrates they broadcasting at.

so... how in the world will encoding regular XM above 64 or 96kbps "preserve" the sound quality?  it won't--- but it'll surely worsen it!  SpasV, although i'm a very technically-minded person myself, i believe that you are ridiculously over-analyzing all of this & making it incredibly more complicated than it actually is... it's not rocket science.  a bit of common sense & two ears will tell you that it's COMPLETELY unnecessary to encode satellite radio shows above what they are broadcast in, especially when you go as high as 320kbps, or God forbid, FLAC!

as it seems you just don't understand my argument & i absolutely don't understand yours, we'll simply have to agree to disagree on this one

OK, let’s go another way.
But first, let me put the problem we are talking about.
So, somewhere two electrical signals, caused by a music sound wave, have been generated. Then these two signals have been transferred through a communication channel, which includes the XM Satellite Radio Cannel 80, which in turns includes a Polk XRt12 XM Reference Tuner. Finally I have a computer file containing the data representing these two original signals – I have two Digital signals.
Let me say, I do not know anything about the communication channel. But I know everything about its output because I have it and I know how to study it.
The Discrete Signal is a Mathematical object – it is a discrete function. I used an application that implements strict mathematical methods developed in the field of Discrete Signal Processing (DSP) theory to process the Digital signals. I received results which I compared with analogous results derived from analogous, but known Digital signals – CD and another sources. Then I compared the results using quantitative and qualitative evaluations and I have reached some conclusions about the Digital signals I have.

Let me take your point now.
It is: You know the communication channel uses a transformation over the signal it carries. You say the transformation is HEC aacPlus v1.
You guess, you “estimate or suppose (something) without sufficient information to be sure of being correct”,  the bit rate “value of xm radio” “alters from 64kbit/s to 96kbit/s”.
In other words you know the signals over the communication channel are first encoded, transferred at some bit rate, and then decoded or reconstructed to be used as sound source.
The implication is: once the signals have been transferred at bit rate of 64 kbits/s the sound quality can not be better than of a sound reconstructed by an arbitrary codec working at the same bit rate. Or in other words it can not be better than those of an mp3 at 64 kbps.
Let me first clarify the meaning of 64/96 kbits/s. I assume it is, obviously the bit rate used by the XM communication channel encoder/decoder.
And here is a question. How much information can be extracted from such bit stream when the sound wave is being reconstructed?
So, how much?
Before that, “Coding Technologies” “is the industry's leading provider of audio compression technologies that offers “a license to patents owned by Coding Technologies and Philips” for using “the Parametric Stereo (PS) technology, e.g. in aacPlus v2”
Here is a reverence from the article “MPEG-4 aacPlus - Audio coding for today’s digital media world”. The authors are Stefan Meltzer and Gerald Moser from Coding Technologies.
“HE-AAC v2 comprises a fully-featured tool set for the coding of audio signals in mono, stereo and multichannel modes (up to 48 channels) – at high quality levels using a wide range of bitrates.
HE-AAC v2 has proven in several independent tests to be the most efficient audio compression scheme available worldwide. The codec’s core components are already in widespread use in a variety of systems and applications where bandwidth limitations are a crucial issue, amongst them:
* XM Satellite Radio – the digital satellite broadcasting service in the USA;
* HD Radio – the terrestrial digital broadcasting system of iBiquity Digital in the USA;
* Digital Radio Mondiale – the international standard for broadcasting in the long-, medium- and short-wave bands.
In Asia, HE-AAC v2 is the mandatory audio codec for the Korean Satellite Digital Multimedia Broadcasting (S-DMB) technology and is optional for Japan’s terrestrial Integrated Services Digital Broadcasting system (ISDB). HE-AAC v2 is also a central element of the 3GPP (3rd Generation Partnership Project) and 3GPP2 specifications and is applied in multiple music download services over 2.5 and 3G mobile communication networks.”

I assume their statement that XM Satellite Radio is amongst the variety of systems that use HE-AAC v2 is true.
The authors refer to a “Multichannel listening tests of the Institut für Rundfunktechnik (IRT)”
According to these tests the authors say “…it can be stated that HE-AAC provides a better quality at half the bitrate compared with WMA or Dolby AC-3.” (Here WMA stands for (Microsoft) Windows Media Audio.)
In other words 64/96 kbits/s is not a bit rate to compare simply as a number with the bit rate of an mp3 encoder. Both the bit rate and the codec determine the amount of information needed when reconstructing a compressed audio signal.
I am not going to assume anything but I would like to suggest that you consider the possibility that the Microsoft WMA codec is not worse than an mp3 codec. If it was true you could conclude that a 96 kbits/s HE-AAC v2 is near equivalent, in sense of audio quality, to 192 kbps mp3.
The next thing I would like to point out is: HE-AAC v2 is a codec family and HE-AAC is a software development kit also. It means there are many applications that implement the HE-AAC v2 codec family. Amongst them:
*Professional encoding products for real-time encoding/streaming and file encoding are available from companies such as Orban, Mayah, and Cube-Tec.”
If you, for example, check out Orban’s “Optimod-FM 8500: Integrated audio processing for the digital radio age” you could find “The 8500's 64 kHz base sample rate allows it to provide up to 20 kHz audio bandwidth at its HD output”
Or, Opticodec-PC 1010 Streaming Encoder - SE Version - Up to 128kbps bitrate using aacPlus codec. Up to 320kbps bitrate using AAC codec.
In other words “HE-AAC v2 bit rate is not a fixed value but as Stefan Meltzer and Gerald Moser say the codec family uses “wide range of bitrates.”
Finally, your arguments are assumptions, and I would say rather free assumptions, and their interpretation is rather free also.
So, what are we talking about?.

A few things:

* Sirius satellite radio is using an earkly AAC audio codec version called PACv2. They broadcast with channel load balancing (an enhanced VBR) at around 48-112kbps (most often between 64-96kbps). The low-pass for Sirius is at 15.4kHz.

* XM satellite radio is using the newer AAC+ with side-band-replication (sbr). That allows them to use the full frequency range up to 20-22kHz (!!!) but the broadcast bitrate is also at 64-96kbps VBR. The higher frequencies beyond 15kHz are artificial (similar to MIDI) because of sbr and no longer original.

* In Europe, satellites are broadcasting in MPEG-I Layer-2 format (MP2). That is (almost) MP3, typical bitrates are 128kbps, 192kbps (low-pass at around 16kHz) and 320kbps.

* In comparison, FM broadcasts are (of course) not bitrate limited but they have a low-pass at around 15kHz, sometimes at 15.4 kHz.


And now a HOW-TO for encoding satellite radio:

* Use LAME! Best LAME encoder settings for all these satellite radios: VBR, VBR and again VBR, without any doubt! VBR takes the available sound information (not really that much!) and uses bitrate and size compression to get the maximum out of it. For low quality sources like those from XM or Sirius the max possible average MP3 bitrate even for "lame -V 0" (the highest possible VBR quality) will be quite low because of that (maybe around 128-160kbps). The quality for LAME -V 0 is almost as high as with FLAC but the size is only a small fraction of the flac file size.

* Recommended Lame encoder settings for all satellite radios: "Lame -V 0 --vbr-new". The frequency cut-off is at your likes but "--lowpass 16" appended to the LAME command-line is probably a good choice (not just for satellite radio but for all broadcasts, even FM). For XM, a low-pass at 16kHz will keep the file size at reasonable values (by removing the space-consuming artificial higher frequencies from sbr).

I would like to add a few things also.
Thanks to the discussion I had to read something about Coding Technologies and their codec family HE-AAC v2 and now I am really interesting in sound compression.
Coding Technologies innovation of Spectral Band Replication could explain to me the spectrum drop around the 16 kHz seen in all XM spectra. So, although the spectrum does not look very good it is not a drawback as I thought but maybe the Spectral Band Replication is not implemented very well.
As it is now clear the world leader in the audio compression through its scientific research has developed the Spectral Band Replication a method for effectively reconstruction of the high band of the spectrum.
Here is what Stefan Meltzer and Gerald Moser say in their article “MPEG-4 aacPlus - Audio coding for today’s digital media world”.
“Spectral Band Replication
In traditional audio coding, a significant amount of information is spent in coding the high frequencies, although the psychoacoustic importance of the last one or two octaves is relatively low. This triggered the basic idea behind SBR. Based on the cognition of a strong correlation between the high- and the low-frequency range of an audio signal (hereafter referred to as the “high band” and the “low band” respectively), a good approximation of the original input signal high band can be achieved by a transposition from the low band (Fig. 4).
Besides pure transposition, the reconstruction of the high band (Fig. 5) is conducted by transmitting guiding information such as the spectral envelope of the original input signal or additional information to compensate for potentially missing high-frequency components. This guiding information is referred to as SBR data. Also, efficient packaging of the SBR data is important to achieve a low data-rate overhead.At the encoder side, the original input signal is analysed, the high band spectral envelope and its characteristics in relation to the low band are encoded and the resulting SBR data is multiplexed with the core coder bit-stream. At the decoder side, firstly the SBR data is de-multiplexed, then the core decoder is used on its own. Finally, the SBR decoder operates on its output signal, using the decoded SBR data to guide the Spectral Band Replication process. A full bandwidth output signal is obtained. Non-SBR decoders would still be able to decode the backwards compatible part of the core decoder ... but resulting in a band-limited output signal only.
Whereas the basic approach seems to be simple, making it work reasonably well in practice is not. Obviously it is a non-trivial task to code the guiding information such that all the following criteria are met:
• Good spectral resolution is required;
• Sufficient time resolution on transients is needed to avoid pre-echoes;
• Cases with non-highly-correlated low band and high band need to be taken care of carefully, since transposition and envelope adjustment alone could sound artificial here;
• A low overhead data-rate is required in order to achieve a significant coding gain.”

The simple conclusion is: The world leader reconstructs the spectrum high band effectively. So, it is worthy to preserve it. So simple, isn’t it?

As to the Lame encoder, I need to collect my observation but the impression I have is that the WavePad (NCH Swift Sound) encoder is better and I use it. Maybe it deserves for me to study the mp3 VBR compression also.

much appreciated! thank you thank you!

@spasv
i highly recommend the lame encoder also. and the method described by Ojay is also a good way to get results what the size should be.
im very happy that Ojay mentioned the artificial high frequency thing i missed to describe till now.
please take the time to do some tests with lame and maybe u can post some examples.

@spasv
i again have to contradict: you dont have a digital data as you would have by ripping a cd or capturing a raw mpa elementary stream from DVB-S by recording from your xm reciever.
because it is sending the true digitized source signal from the satellite tru the propriatery DSP of the reciever. and this deliviers you an PCM like waveform that in return u can record via the line out or S/PDIF