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: Accuracy FLAC decoder (Read 28536 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Accuracy FLAC decoder

Hello! I often hear the opinion that the need to trans-code FLAC files to WAV to achieve the highest quality sound. While about WV files or APE that did not say. I use a 16-bit PC sound cards and my subjective opinion, at low frequencies there is some advantage APE over the FLAC. Is it possible to learn more about how it works FLAC-decoder step-by-step? Library to compile the decoder, is only one, or if there are several decoders FLAC? Is it possible (not in the laboratory) to objectively compare Decoders APE and FLAC for instance?

Accuracy FLAC decoder

Reply #1 –
Quote from: Bublic on
Hello! I often hear the opinion that the need to trans-code FLAC files to WAV to achieve the highest quality sound. While about WV files or APE that did not say. I use a 16-bit PC sound cards and my subjective opinion, at low frequencies there is some advantage APE over the FLAC. Is it possible to learn more about how it works FLAC-decoder step-by-step? Library to compile the decoder, is only one, or if there are several decoders FLAC? Is it possible (not in the laboratory) to objectively compare Decoders APE and FLAC for instance?


FLAC, APE, TAK and other lossless codecs are LOSSLESS. There is no sound quality difference what so ever. Those who believe otherwise is ignorant.
What lossless means: If you decode the compressed file you will get the exact same result as the original file; the WAV. It will even have the same checksum.
Try this yourself in Foobar, with the use of this "Bit Compare" component: https://www.foobar2000.org/components/view/foo_bitcompare
If you still don't believe it, try to ABX a WAV from lossless compressed version of it using this component: https://www.foobar2000.org/components/view/foo_abx

One more thing: When playing a lossless (or even lossy) compressed file, most if not all players decode it to WAV PCM. For instance, you can't play a FLAC file without a decoder, which decodes FLAC to WAV.

Accuracy FLAC decoder

Reply #2 –
o-l-a-v
But you can't deny the fact that the WV, APE and remaining decoders load CPU intensive greater than FLAC? This fact has two logical explanation:
1. Other codecs poorly optimized.
2. Decoder FLAC simplifies and thus impairs processing decompressed file.

Accuracy FLAC decoder

Reply #3 –
Quote from: Bublic on
Hello! I often hear the opinion that the need to trans-code FLAC files to WAV to achieve the highest quality sound. While about WV files or APE that did not say.

Probably that's because FLAC is more popular. Which means that it is also more popular among audiophools.


Quote from: Bublic on
I use a 16-bit PC sound cards and my subjective opinion, at low frequencies there is some advantage APE over the FLAC.

Just imagination. Only randomized double-blind comparisons are valid.


Quote from: Bublic on
Is it possible to learn more about how it works FLAC-decoder step-by-step?

I don't know if it'll help you, but FLAC format description is available: https://xiph.org/flac/format.html


Quote from: Bublic on
Is it possible (not in the laboratory) to objectively compare Decoders APE and FLAC for instance?

They produce identical results, so there's nothing to study.

Accuracy FLAC decoder

Reply #4 –
Quote from: Bublic on
But you can't deny the fact that the WV, APE and remaining decoders load CPU intensive greater than FLAC? This fact has two logical explanation:
1. Other codecs poorly optimized.
2. Decoder FLAC simplifies and thus impairs processing decompressed file.


That's plain stupid. FLAC was designed for fast decoding.
FLAC decoder cannot "simplify" decoding because FLAC is a lossless codec.

Accuracy FLAC decoder

Reply #5 –
There is no evidence that high system load negatively affects sound quality.

Anyway, FLAC was designed for very fast decoding. According to xiph.org it is one of the if not the fastest lossless decoders. APE is much slower (higher load).

But speed has no impact on sound quality, since we're talking LOSSLESS here. More complex codecs just require more calculations (more time, higher load) on the CPU to arrive at the SAME result.
"I hear it when I see it."

Accuracy FLAC decoder

Reply #6 –
Speed may make a difference in battery life with portable players, well that's one part of the problem with those things with regards to battery life.  On slower systems FLAC at level 8 maybe playable compared to Monkey's Audio being set to Insane which may not be playable.

All lossless decoders arrive at the same result otherwise it wouldn't be lossless.

Accuracy FLAC decoder

Reply #7 –
Quote from: Bublic on
o-l-a-v
But you can't deny the fact that the WV, APE and remaining decoders load CPU intensive greater than FLAC?


WV is pretty close to FLAC. Both are designed for CPU/power efficient decoding.

Look up what lossless means and save everyone (and yourself) a lot of wasted time.

Accuracy FLAC decoder

Reply #8 –
The problem is not so much in the encoder as the plugin output players software.
Winamp FLAC (16/44.1):


Winamp WAV (16/44.1):


It is evident that Winamp out_wave.dll has the correct 16-bit output to WAV, and the 24-bit for 16 / 44.1 flac, which leads to unnecessary conversion 16 - 24 - 16 in the case of fair 16-bit DAC. (Files APE unfortunately could not get to play using Winamp.) And with all certainty I do not know what would actually have FLAC, APE, WAV concluded foobar2000, for lack of diagnostic information in the plugin foo_out_ks or system ASIO.

Accuracy FLAC decoder

Reply #9 –
Quote from: Bublic on
It is evident that Winamp out_wave.dll has the correct 16-bit output to WAV, and the 24-bit for 16 / 44.1 flac, which leads to unnecessary conversion 16 - 24 - 16 in the case of fair 16-bit DAC.


This is not a problem unless you are dealing with >24 bit files. 

Accuracy FLAC decoder

Reply #10 –
Quote from: Bublic on
It is evident that Winamp out_wave.dll has the correct 16-bit output to WAV, and the 24-bit for 16 / 44.1 flac, which leads to unnecessary conversion 16 - 24 - 16 in the case of fair 16-bit DAC.

You can uncheck 'Allow 24 bit' option in General Preferences -> Playback. But as saratoga said it's not a problem. By the way, why you use outdated WaveOut instead of DirectSound?

Quote from: Bublic on
Files APE unfortunately could not get to play using Winamp.

You need the Monkey's Audio plugin

Quote from: Bublic on
And with all certainty I do not know what would actually have FLAC, APE, WAV concluded foobar2000, for lack of diagnostic information in the plugin foo_out_ks or system ASIO.

Also doesn't matter. I don't think that ASIO or KS can improve audio quality.

Accuracy FLAC decoder

Reply #11 –
Quote from: lvqcl on
You can uncheck 'Allow 24 bit' option in General Preferences -> Playback. But as saratoga said it's not a problem. By the way, why you use outdated WaveOut instead of DirectSound?

Yes I can. But then 24/96 converted into 16/96 in Winamp, and not the fact that this conversion will be done in the best possible way. I have more confidence in the algorithm that is provided by the DAC.
Perhaps these settings should be Winamp for default, I do not know because I do not usually use this player. For DS the same picture as in the above figures. The question is, what happens to output sound in foobar2000. If the same as in Winamp, then, obviously, it's bad.
Quote from: lvqcl on
You need the Monkey's Audio plugin

It does not work, unfortunately.

A Monkey's Audio, if I'm not mistaken, in principle does not support 24-bits, and may therefore preferable than FLAC. And in this whole thing.

Accuracy FLAC decoder

Reply #13 –
Quote from: Bublic on
Yes I can. But then 24/96 converted into 16/96 in Winamp, and not the fact that this conversion will be done in the best possible way. I have more confidence in the algorithm that is provided by the DAC.

If your DAC supports only 16bit then 24->16 bit conversion is made before DAC: in a player, driver or Windows audio stack.

Quote from: Bublic on
The question is, what happens to output sound in foobar2000. If the same as in Winamp, then, obviously, it's bad.

"obviously"... I don't think there's anything bad in it. Maybe it's less than ideal, but... not everyone sets volume slider to 100%.

Quote from: Bublic on
It does not work, unfortunately.

Works here. Requires WinXP(+maybe some service pack).

Quote from: Bublic on
A Monkey's Audio, if I'm not mistaken, in principle does not support 24-bits

It does support 24 bit. But maybe its plugin won't convert 16-bit audio to 24-bit.

Accuracy FLAC decoder

Reply #14 –
This doesn't go anywhere.

@Bublic: you said that FLAC is faster or requires less CPU on your computer than the other lossless codecs. That means that even with the possible 16 bit to 24 bit conversions, this is still the most efficient route compared to anything else. So there is no question there.

You don't like the idea of having bit depth conversions because of it could potentially degrade the audio? I understand your concern. However that's what ABX tests are for and you will have to take one before going further here. Further groundless discussion is futile.

As a final note, I guess you are aware that most, if not all, audio pipelines include many other bit depth conversions: most filters, from equalizers to efficient low-pass filters and volume change generally work best with floating point numbers for instance. I wouldn't be surprised if Windows converted everything to float32. The thing is that this is a very good thing: the relative round off errors of 16 bit ints are far greater than the one of 32 bit floats.

Accuracy FLAC decoder

Reply #15 –
MMime  I never was a parishioner of the Church the Holy DBT, so I will refer to the closest analogue. You can certainly distinguish between 8-bit and 16-bit audio signals. If yes, apparently it is also possible to distinguish between themselves and others, 16-bit to 24-bit. Simply search using the proposed tool ABX (DBT) is difficult to do. However, proposed by this analogy no longer be less accurate.
You do not pay attention to what I have argued from the outset that I consider better sound format APE, and apparently the feeling found an objective and unexpected for me to explain. More more fully ascertain the validity of this conclusion does not prevent the lack of ABX test, and the lack of logs at the output plugins in foobar2000, I think so.

Accuracy FLAC decoder

Reply #16 –
Could you post the winamp file info window for the FLAC file please.
Who are you and how did you get in here ?
I'm a locksmith, I'm a locksmith.

Accuracy FLAC decoder

Reply #18 –
Quote from: Chibisteven on
Speed may make a difference in battery life with portable players, well that's one part of the problem with those things with regards to battery life.  On slower systems FLAC at level 8 maybe playable compared to Monkey's Audio being set to Insane which may not be playable.

All lossless decoders arrive at the same result otherwise it wouldn't be lossless.


Even that maybe uncertain with current hardware . My ancient deprecated legacy GTS5830 can play APE fine except for the highest modes. There may not be any battery difference using APE fast setting vs flac , Yet APE [fast] is without question more efficient than flac -8. When where these formats [ape,wv,flac] created ?  Late 90's.  Their default settings were fine on PC hardware of 1999.

Accuracy FLAC decoder

Reply #19 –
Quote from: Bublic on
It is evident that Winamp out_wave.dll has the correct 16-bit output to WAV, and the 24-bit for 16 / 44.1 flac, which leads to unnecessary conversion 16 - 24 - 16 in the case of fair 16-bit DAC.
If you disable dithering, this is essentially a no-op. What exactly are you worrying about?

Quote from: Bublic on
MMime  I never was a parishioner of the Church the Holy DBT, so I will refer to the closest analogue. You can certainly distinguish between 8-bit and 16-bit audio signals. If yes, apparently it is also possible to distinguish between themselves and others, 16-bit to 24-bit.
No. That doesn't follow. You can only measure as good as your measurement equipment allows it, and it has not been demonstrated that human brains and ears are good enough to distinguish between 16 and 24 bit audio reliably.

Quote from: Bublic on
Simply search using the proposed tool ABX (DBT) is difficult to do. However, proposed by this analogy no longer be less accurate.
You do not pay attention to what I have argued from the outset that I consider better sound format APE, and apparently the feeling found an objective and unexpected for me to explain. More more fully ascertain the validity of this conclusion does not prevent the lack of ABX test, and the lack of logs at the output plugins in foobar2000, I think so.
This is not understandable. Can you try to rephrase that?

I wonder what the actual topic is now. Is it still about lossless formats? Then the answer is lossless is lossless and any other discussion is moot. Is it about how to set up Winamp to do as little processing as possible?
It's only audiophile if it's inconvenient.

Accuracy FLAC decoder

Reply #20 –
Quote from: Kohlrabi on
I wonder what the actual topic is now. Is it still about lossless formats? Then the answer is lossless is lossless and any other discussion is moot. Is it about how to set up Winamp to do as little processing as possible?


As it's really about the warm and fuzzy feeling one gets when things are set up as "perfectly" as they could ever be, regardless of audibility, based on an incomplete understanding of difference and significance, it's about both.
Creature of habit.

Accuracy FLAC decoder

Reply #21 –
Random value 22483 as 16 bit word in binary, then left shifted by 8 bits to a 24 bit word, then left shifted again to a 32 bit integer, then right shifted again into a 16 bit word:
Code: [Select]
1010111 11010011
1010111 11010011 00000000
1010111 11010011 00000000 00000000
1010111 11010011


These are all lossless operations, because they effectively don't do anything.
"I hear it when I see it."

Accuracy FLAC decoder

Reply #22 –
Quote from: Bublic on
I never was a parishioner of the Church the Holy DBT, so I will refer to the closest analogue. You can certainly distinguish between 8-bit and 16-bit audio signals. If yes, apparently it is also possible to distinguish between themselves and others, 16-bit to 24-bit.

You probably can distinguish 1.1 from 1.2 kilograms, bot it doesn't mean that you can distinguish 1.001 from 1.002 kg (without measurements).

Accuracy FLAC decoder

Reply #23 –
Kohlrabi
I'm sorry. Usually in this discussion, I mean lossless, but lossy formats although subjected to recalculation in 24-bit. And yet discussed the topic of audio mastering - I think that here it would be... redundant?
In order not to frighten anyone, I must say that on the surface there is a problem only for those, who use 16-bit DACs. There are not many among the total, but they still exist.
If you are reading this topic, there is a standard (24-bit) PC sound card, then you have this topic should not disturb!
(Winamp also sufficiently good, as for example when coding your own fhgaac encoder 128 aac sbr it shows excellent quality results, imho.)

lvqcl
Quote from: lvqcl on
You probably can distinguish 1.1 from 1.2 kilograms, bot it doesn't mean that you can distinguish 1.001 from 1.002 kg (without measurements).

Ok, ok, I'm not a jeweler.

Accuracy FLAC decoder

Reply #24 –
Quote from: Bublic on
In order not to frighten anyone, I must say that on the surface there is a problem only for those, who use 16-bit DACs. There are not many among the total, but they still exist.


This is not a problem for anyone.  Not people with 16 bit DACs, or 24 bit or any other number.  It simply does not matter. 

You are worrying about nothing at all.