• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Converting WAV to FLAC which FLAC Level is sufficient

digicidal

Major Contributor
Joined
Jul 6, 2019
Messages
1,981
Likes
4,838
Location
Sin City, NV
I can possibly understand the assertion, if the problem lies elsewhere in the process... but not with the files themselves (as pertaining to FLAC.. not the MP3 part that somehow got into this).

@maty If your PC is remote from the playback device and you aren't hearing noise generated by the device itself - then it's surely imagined. I'm very sensitive to coil whine and can hear the MB, PSU & GPU when they are heavily loaded (gaming, video processing, 3D renders, etc) - and in certain configurations have heard it feedback through my monitors. It's like a choir of mosquitos practicing for a concert.

However, it wasn't degraded audio quality of the source material - it was additional noise over the top of it. However, regardless of the compression level, nothing decoding-wise should ever be loading your system enough to matter unless you're doing something else far more stressful at the same time.

When pulled directly from the NAS over the LAN to a streaming DAC... I know, blind or sighted, there isn't an audible difference. But it's easy enough to just capture the audio you believe is very different and analyze the result. If the waveform is the same... you're making up the difference, and if it's different you might not be - depending on the significance of that difference.

Regardless, if you can hear a difference between different levels of FLAC compression - you have a problem with your software or implementation... not with the format. That's why you can convert to WAV again and have a bit perfect version of the source. I wonder if there could be a problem with your ADC used to capture the original WAV from vinyl?

Does anyone have a logical explanation about why the MP3 at 48 kHz sounds much better than 44 kHz?
The difference is very noticeable, it misplaces you.

Unless there are phase issues that were induced that shouldn't be the case (unless I'm misunderstanding "misplaces you").
 
Last edited:

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
The PC components were chosen one by one, including the box. One of the objectives was to be silent. It doesn't even have an external graphics card, which I don't need because I don't play or render large videos. The SMPS source, silent, also has little curl at +5 Vdc.

Blessed problem that has been with me for years, with two different computers and CPU and two windows (7 and 10). with a few Linux tested, including some designed for audio, the sound quality was lower, with the different computers.

The expensive ADC and the vinyl rip are not the problem. I have detected the differences for many years (vinyl, CD, SACD usually), as I have already written.

MP3. I did not expect that big sound difference! I believed that the psychoacoustic algorithm would match the sound. Other day, in another test, I discovered that the MP3 generated from 24/96 FLAC or WAV was at 48 kHz and not at 44 kHz, hence its incorporation into the test. To be at 44 kHz I have to do the resampled first from 96 kHz to 44 kHz FLAC or WAV.
 

JJB70

Major Contributor
Forum Donor
Joined
Aug 17, 2018
Messages
2,905
Likes
6,148
Location
Singapore
Last night I finally got around to ripping the EMI/Warner Celibidache Bruckner set to FLAC. As usual I used a cheap USB CD drive (£15) and my Microsoft Surface Pro tablet with Windows media player. So nothing exotic or which is not bundled software forming part of Windows. If there are any problems with the result then discerning them is beyond my capabilities.
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,499
Likes
5,417
Location
UK
Does anyone have a logical explanation about why the MP3 at 48 kHz sounds much better than 44 kHz?
The difference is very noticeable, it misplaces you.
Yes, you did a test that was not double blind, and you have not started that the levels were matched.
Do some controlled listening tests or shut up, you just degrade the signal to noise on this forum with unscientific nonsense.
 

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
Celibidache/Bruckner. Only I have CD. Celibidache, Munchner P - Bruckner - Symphony No 4 (1993), Symphony No 5 [Berlin] (1986) and Symphony No 5 [Tokio] (1986).

It is clear that the volume is always the same during the test. The sound was more open, clear, more spatial at 48 kHz.

Anyway, nobody asks me for the files :oops:
 

graz_lag

Major Contributor
Joined
Oct 13, 2018
Messages
1,296
Likes
1,583
Location
Le Mans, France
... If there are any problems with the result then discerning them is beyond my capabilities ...

Indeed, you did not perceive any differences because there are no audible differences between WAV and FLAC as demonstrated by 'thousands' of ABX tests ...
Even the file comparing command-line 'diff' do not see any difference in the music content.
 

JJB70

Major Contributor
Forum Donor
Joined
Aug 17, 2018
Messages
2,905
Likes
6,148
Location
Singapore
The only reason I use FLAC instead of 320k MP3 is because memory is cheap and I am playing the usual psychological games in telling myself it is better. I can differentiate some FLAC and MP3 files but not all and not reliably. And the concentration required is not much fun. Which tells me that for just listening to music and enjoying the experience MP3 is fine.
 

graz_lag

Major Contributor
Joined
Oct 13, 2018
Messages
1,296
Likes
1,583
Location
Le Mans, France
Celibidache/Bruckner. Only I have CD. Celibidache, Munchner P - Bruckner - Symphony No 4 (1993), Symphony No 5 [Berlin] (1986) and Symphony No 5 [Tokio] (1986).

It is clear that the volume is always the same during the test. The sound was more open, clear, more spatial at 48 kHz.

Anyway, nobody asks me for the files :oops:

Why are you so stubborn abt. the 48kHz format ? Unless you intend to use it for film/video editing.
What is your source sample rate ?
I am a very simple guy : I keep @ 48kHz a source recorded @ 48 or 96, as with the same approach, @ 44.1 or 88.2 a source recorded @ 44.1kHz.
Why switching between 48 and 44.1 ? What do I know abt. the resamplig algorithm of my converting software ?

It's math, for a given data rate, a 44.1kHz mp3 is of higher quality than a 48kHz mp3, but - as we all know, the human ear could not hear the difference between the two. Picking the right audio sample is simply a matter of luck : 50/50 or whatever other combination you like.
 

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
It was an accident. I almost never listen to MP3. For the phone I convert from FLAC to Ogg Vorbis -q9 to save space. I assumed that the conversion from 24/96 would be automatically to 16/44, as with the 16 bits. With LAME 3.99r encoder.

FLAC vs WAV test will also be done from Gnu/Linux Manjaro. And I will listen to both MP3.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
44.1 kHz multiples will be converted to 44.1 kHz in MP3.
48kHz multiples will be converted in 48kHz MP3.
Some encoders even retain 24 bit depth (AFAIK) but most decoders/players truncate to 16 bits on playback.
 

infinitesymphony

Major Contributor
Joined
Nov 21, 2018
Messages
1,072
Likes
1,806
Yes, converting to a lossy format and resampling may sound better to some people. Objectively speaking, it is not better. Does it play with the frequency response and spatial relationships? Sure.
 

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
* Gnu/Linux Manjaro XFCE 18.0.4 64 bits - Audacious player

A 50ms buffer is not possible, only 100ms. Tested with PulseAudio and ALSA output.

Manjaro-xfce-18.0.4-stable-x86_64-Audacious-aqualung-test.jpg


I have barely noticed the difference between the four files. I think 24/96 sounded a little better than 24/44 but it would never have surpassed an ABX. Very good sound in my second audio system, near field.


* Windows 10 1809 64 bits - foobar2000 1.4.2. Without win optimizacion to play multimedia. Buffer 50ms, Kernel Streaming. Hard to beat an ABX among the four. I notice the difference, subtle, of MP3 16/48 compared to FLAC but not with 16/44.

foobar2000-KS-aqualung-test.jpg


I like more the sound from Manjaro/Audacious! I have been complaining about the foobar2000 sound for many years and not only with the ABX plugin.


Note: Artist like 123 to find files quickly from Gizmo app.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
Manjaro is a righteous linux. Audacious is my go to player. Clementine if I want all the extra features, but for just lining them up and playing Audacious rocks.
 

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
Curious, with the first song:

* Audacious: 96 kHz, 3051 kbps

* foobar2000: 96 kHz, 3048 kbps
 

infinitesymphony

Major Contributor
Joined
Nov 21, 2018
Messages
1,072
Likes
1,806
Probably just a difference in bitrate estimation.

Around 10 years ago, upsampling players and DACs were en vogue. Now they're not. You are welcome to do whatever you like to increase the enjoyment of your music; use tube amplifiers, resample, decimate, excite, compress/limit/expand, EQ, you name it. However, none of those modifications will bring you closer to the source than the original unmodified PCM or DSD.
I like more the sound from Manjaro/Audacious! I have been complaining about the foobar2000 sound for many years and not only with the ABX plugin.
If you believe there is a difference, you should digitally record the output of the same file from both applications and use SoX to show the differences:

https://sound.stackexchange.com/que...wo-similar-audio-files-using-graphical-method

If you can find a difference, report it to the developers.
 

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
The difference it was very obvious. I do not like foobar2000 as a player. Then I listened again on 24/96 with the optimized win. It improved in foobar2000 but I was not convinced by the sound. I went to JRiver MC and it was where it sounded the best.

In Audacious there were too much bass, of which I was aware later. At first that pleases but covered details.

It seems that the differences lie in the playback software and the optimization I do to windows. Optimization that allows me to differentiate between files and easily pass some tests proposed by PMA in diyaudio.com.
 

infinitesymphony

Major Contributor
Joined
Nov 21, 2018
Messages
1,072
Likes
1,806
You may want to consult your signature and go looking for more facts before making proclamations that you have made no objective attempt to verify.
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,260
Likes
16,298
Location
uk, taunton
@maty your taking us into your maddening black hole again and at the usual frantic pace.

Little your write in terms of the certainties attached to your subjective listening tests mean anything to anyone here other than you and are certainly not the basis for protracted discussion.

I know it fascinates you and I appreciate its a passion of yours but this is just not the place for this kind of thing.
 
Top Bottom