Converting WAV to FLAC which FLAC Level is sufficient

[bennetng]

 

[Veri]

My experience contradicts my/our theoretical knowledge but it is what it is. What do I gain by publicly defending an issue that goes against what most people experience? And that also I can not show with measurements. It is frustratingly and irritating to me and others.

Maybe @amirm can verify it

 

[maty]

View attachment 27916

When disqualification is used as an argument, whoever disqualifies is the author. As I have written, it is irritating that my theoretical knowledge does not match with my listening experience. Having written it so recently, that image is not funny.

 

[SIY]

When disqualification is used as an argument, whoever disqualifies is the author. As I have written, it is irritating that my theoretical knowledge does not match with my sighted listening experience. Having written it so recently, that image is not funny.

Fixed it for you.

 

[simbloke]

You've described close to nothing about what your actually doing or what you are using. There's nothing for people to believe.

 

[maty]

Unfortunately, misconduct has been generalized in the open Internet, and this forum is no exception.

 

[flipflop]

As I have written, it is irritating that my theoretical knowledge does not match with my listening experience.

The root of the issue is that the concept of cognitive biases is not a part of your theoretical knowledge as you have demonstrated time and time again.

 

[bennetng]

that image is not funny

But UFO is real. You don't believe me?

 

[Frank Dernie]

My experience contradicts my/our theoretical knowledge but it is what it is. What do I gain by publicly defending an issue that goes against what most people experience? And that also I can not show with measurements. It is frustratingly and irritating to me and others.

The placebo effect is powerful and real. If you believe you can hear a difference, you will, even when there is none.

 

[Vincent Kars]

Hi Maty
Just a little bit more FLAK
You claim is “ that I am able to differentiate between different degrees of compression”
You also claim “WAV vs FLAC 0 I am unable to differentiate, as expected”

WAV compresses 0%, FLAC 0 compresses approximately 70%, FLAC 5 approximately 65%
https://web.archive.org/web/20090202063734/http://synthetic-soul.co.uk:80/comparison/lossless/index.asp
I would say that if you are able to hear the difference between 70% and 65%, you must be able to hear the difference between 0% (WAV) and FLAC 0 (70%) as well
Or do you belong to those who mistake FLAC 0 for zero compression?
Ever tried FLAC without compression?

-0 --disable-fixed-subframes --disable-constant-subframes

 

[zermak]

My experience contradicts my/our theoretical knowledge but it is what it is. What do I gain by publicly defending an issue that goes against what most people experience? And that also I can not show with measurements. It is frustratingly and irritating to me and others.

I don't want to be rude or anything but if you can't measure it somehow and we have plenty of instruments to measure audio gear and so on now then I remain skeptical.

Hi Maty
Just a little bit more FLAK
You claim is “ that I am able to differentiate between different degrees of compression”
You also claim “WAV vs FLAC 0 I am unable to differentiate, as expected”

WAV compresses 0%, FLAC 0 compresses approximately 70%, FLAC 5 approximately 65%
https://web.archive.org/web/20090202063734/http://synthetic-soul.co.uk:80/comparison/lossless/index.asp
I would say that if you are able to hear the difference between 70% and 65%, you must be able to hear the difference between 0% (WAV) and FLAC 0 (70%) as well
Or do you belong to those who mistake FLAC 0 for zero compression?
Ever tried FLAC without compression?

-0 --disable-fixed-subframes --disable-constant-subframes

Maybe he is able to spot differences in BMP and PNG images also

 

[bennetng]

differences in BMP and PNG

With transparency layer I can spot the difference

 

[mansr]

FLAC uses linear prediction.
The advantage is that the prediction can be written as a couple of coefficients. They can be stored using very little space. The bad news is that there is nothing linear about audio. Hence there will be a residue. Store the coefficients + residue in a file and the decoder can reconstruct the original signal.

The “compression” level of FLAC is the amount of resources FLAC is allowed to spend on finding the best possible linear prediction. Obvious, the better this prediction, the smaller the residue.
However this will affect most of all the amount of CPU used when encoding.
At decoding time (playback), there is no search needed for an optimal solution. The decoder simply calculates the linear prediction using the coefficients and add the residue to it.
As FLAC uses FIR Linear prediction for its highest compression level, there is still a small performance penalty at playback time compared with e.g. level 0.

So a higher compression level will most of all affect encoding time and to a far lesser extend affect decoding time.
On todays CPU’s you will probably have a hard time to measure the difference between level 0 and level 8 even when encoding.
I/O I expect to be the limiting factor, not the CPU.

Moreover, the effort required to parse the compressed bitstream is roughly proportional to the bit rate regardless of what happens afterwards. A file compressed to a higher ratio can easily be cheaper to decode than a less compressed equivalent.

 

[zermak]

With transparency layer I can spot the difference

Ahahah. And if that is serious tho (because you could considering that indeed BMP hasn't it) I was talking about BMP (raw informations) to PNG compression/conversion (like WAV to FLAC ).

@Vincent Kars why did they implement a non compression option on FLAC? What is the point in doing it? Just from a studying point of view?

 

[bennetng]

Maybe @amirm can verify it

If decoding a flac file is such a demanding task, then all Amir's measurements are invalid, just imagine how the AP software suite running complex FFT and such during measurement.

 

[mansr]

@Vincent Kars why did they implement a non compression option on FLAC? What is the point in doing it? Just from a studying point of view?

Almost all compression formats include the option to store a chunk of data verbatim. The reason for this is that there will always exists inputs that the chosen algorithm can't compress. When such a block is encountered, the most efficient action is to simply store it unaltered since applying the coding method would actually make it larger. This is an unavoidable mathematical fact in any lossless compression scheme.

 

[zermak]

If decoding a flac file is such a demanding task, then all Amir's measurements are invalid, just imagine how the AP software suite running complex FFT and such during measurement.

Plus we have to consider that the measurements provided by @Vincent Kars are on a prehistoric Athlon XP 2400+ (it doesn't even have SSE2 instuctions set support!) and its single core performance are way behind the ones of a modern processor in our systems and maybe even the storage system (see mechanical HD) was a bottle neck for the actual performance of it...

 

[maty]

Hi Maty
Just a little bit more FLAK
You claim is “ that I am able to differentiate between different degrees of compression”
You also claim “WAV vs FLAC 0 I am unable to differentiate, as expected”

WAV compresses 0%, FLAC 0 compresses approximately 70%, FLAC 5 approximately 65%
https://web.archive.org/web/20090202063734/http://synthetic-soul.co.uk:80/comparison/lossless/index.asp
I would say that if you are able to hear the difference between 70% and 65%, you must be able to hear the difference between 0% (WAV) and FLAC 0 (70%) as well
Or do you belong to those who mistake FLAC 0 for zero compression?
Ever tried FLAC without compression?

-0 --disable-fixed-subframes --disable-constant-subframes

As the compression is reduced the harder it was to appreciate the difference. That is, it is easier with FLAC 8 than with FLAC 4, hence I expected it to be more difficult or impossible with FLAC 0.

 

[Vincent Kars]

on a prehistoric Athlon XP 2400+

Yep, there are reasons why Josh Coalson added this "compression" effort parameter

 

[Vincent Kars]

Almost all compression formats include the option to store a chunk of data verbatim. etc

Spot on.

FLAC uses four methods for modeling the input signal:

• Verbatim. This is essentially a zero-order predictor of the signal. The predicted signal is zero, meaning the residual is the signal itself, and the compression is zero. This is the baseline against which the other predictors are measured. If you feed random data to the encoder, the verbatim predictor will probably be used for every subblock. Since the raw signal is not actually passed through the residual coding stage (it is added to the stream 'verbatim'), the encoding results will not be the same as a zero-order linear predictor.
• Constant. This predictor is used whenever the subblock is pure DC ("digital silence"), i.e. a constant value throughout. The signal is run-length encoded and added to the stream.
• Fixed linear predictor. FLAC uses a class of computationally-efficient fixed linear predictors (for a good description, see audiopak and shorten). FLAC adds a fourth-order predictor to the zero-to-third-order predictors used by Shorten. Since the predictors are fixed, the predictor order is the only parameter that needs to be stored in the compressed stream. The error signal is then passed to the residual coder.
• FIR Linear prediction. For more accurate modeling (at a cost of slower encoding), FLAC supports up to 32nd order FIR linear prediction (again, for information on linear prediction, see audiopak and shorten). The reference encoder uses the Levinson-Durbin method for calculating the LPC coefficients from the autocorrelation coefficients, and the coefficients are quantized before computing the residual. Whereas encoders such as Shorten used a fixed quantization for the entire input, FLAC allows the quantized coefficient precision to vary from subframe to subframe. The FLAC reference encoder estimates the optimal precision to use based on the block size and dynamic range of the original signal.

From https://xiph.org/flac/format.html#prediction