I mostly download bought albums in AIFF. AIFF takes even more space vs WAV, but I prefer AIFF.
Wav sound a bit flat. Flac more airy. AIFF fuller vs Flac, but not flat.
I know, there’s no logic. But I repeated the comparison on different days, different moods, with my Melco settled on Flac and on Wav.
Agreed.
That’s why I said above that there must be an explanation of why a user prefers one to the other - but we don’t know what that is.
And we probably never will.
Sure, but as a technical person I am interested in, or at least intrigued by, the why, in particular if it’s the opposite of what had been insisted on for years
The places I’d look for a technical explanation would include:
Exactly how are the data stored on arrival?
How are they pulled out of storage and fed into the DAC?
How is the timing of the data handled in the digital to analog conversion?
How (if at all) could the differences between the 2 processes make an audible difference to SQ?
None of this has much to do with the FLAC. When the DAC yes the data, it sees the audio data samples.
But whatever, it only came up because FR says he prefers a certain lossless file format over another, and it’s the opposite to what the people with golden digital ears have insisted on for years. I mainly found it funny because of the past discussions between myself and FR, all in good mood, and a bit intriguing.
Found another member like me, preferring FLAC.
“ Richieroo
I have tried both … to me Naim seems to sound a little better with flac … I also don’t particularly like DSD on Naim… Flac is superb…even std 16bit if recorded well is amazing…”
I fully believe that people randomly prefer one or the other, and would do so even in cases where there is objectively no difference*. That doesn’t tell us much either way, unfortunately.
It’s probably best if we leave it there
Edit: To be clear, it doesn’t mean that there is no difference, either, just that it’s super difficult to find out either way - as always, so the discussion can continue forever
* and there’s much evidence for this
I understood earlier that you didn’t understood how could I prefer FLAC VS WAV, finding my preference going against logic. You even added a .
It’s the reason we have this discussion.
I would add another one, just for the fun:
“ GraemeH
1
I have FLAC & WAV versions of the same rip on my NAS as when I moved from HDX years ago I got it to convert them in preparation for using an NDX.
After a bit of tooing and froing testing between both I prefer the FLAC on the NDX2.”
I wasn’t laughing about you, sorry if this was not clear. I just found it innocently funny because we have had so many similar discussions. And in this particular WAV vs FLAC thing, I was sceptical for a long time for why people would prefer transcoding to WAV, but eventually I have agreed for some time that the decompression work for FLAC is not an entirely bad explanation in some situations. And when I come around to this view, you then prefer FLAC. It’s just funny.
I believe that most people agree that on the new streamers the difference is negligible, and it makes sense because they have much more computing power than the legacy streamers, so the minimal load from FLAC decompression should not be stressing them.
And human perception being as unreliable as it undoubtedly is, it’s not surprising that our brain invents differences and patterns when there are none. It is how our brain was wired by evolution, and we can see this in action everywhere. It is precisely why humankind had to come up with the convoluted scientific method, because without it our brains are not reliable.
Scanned much of this discussion. Clearly the only way to resolve this and all the other controversial issues, network cables, switches etc…etc. would be to conduct blinded appropriately designed trials.
Tedious for sure but is the only way to get the correct answer.
In my view, the problem with this is that to be statistically significant you need so many test subjects and such a complex test setup that it’s not practical. And in a highly subjective topic like this one, the mere act of asking has an influence on the subject.
I’ve mentioned it in another, older post about another blind test discussion: A German TV station once did a blind test in a TV show with Anne-Sophie Mutter, a conductor, and a violin professor (or something). They were asked to distinguish a Stradivarius, a modern world class concert violin, and a good beginner’s violin. And they failed. I do not believe that these test subjects have cloth ears, or that there are no differences between these violins. I choose to believe that the test setup was flawed, and blind tests of hearing are difficult
Thanks for clarifying. I know you were not directly laughing at me.
Sometimes it’s better leave without understanding, and just believe our ears.
I sometimes thought myself’ perhaps should I transcode to WAV, as the majority agree on that “. But then my intuition replicated “ no, it’s stupid, better leave what I prefer to listen to “.
Correct me if I am wrong I think in order for a Naim streamer to play dsd it converts it back to flac…hence an extra process…hence quality impact…???
Hi Richieroo, its not back to FLAC,
As I understand it … In Naim streamers the DSD signal is transcoded to PCM in the DSP and is low-pass filtered to remove DSD’s noise-shaped ultrasonic quantisation noise before passing to the DAC stage.
Another method used by some systems that have a DAC enabled for is DoP (DSD over PCM). DoP is DSD data packaged within stereo 24 PCM data framing, there is no conversion, DoP is simply a means of streaming unmodified DSD & native DSD is maintained end to end.
Cheers…not very technical…I knew it converted it…
What about MQA v FLAC v WAV anyone……
Different things. MQA is not a file format like FLAC or WAV, and not a lossless compression algorithm like the one that is part of the FLAC specification, but a lossy one. The MQA-encoded data can then be embedded in FLAC or ALAC container files. The compression embeds the low-energy high-frequency data band into the low-frequency band, saving space. If played by non-MQA equipment, it usually only leaves 13 bits of PCM music data, instead of 16. (The lower (least significant) 3 bits are rendered as noise by non-MQA equipment)
Hi, correct, FLAC is a lossless compressed format encoding PCM data. Wav files are, usually, a lossless non compressed format encoding PCM data.
Both formats need to read or decoded to construct a timed data PCM feed to go to a DAC either via USB or SPDIF typically.
As you say FLAC to WAV is not automatic … it can be done in software such as in a UPnP media server for example.
There is also the matter of metadata. FLAC contains Vorbis and ID3 tags… and was optimised to suit ripping and consumer track info.
Wav meta data originally was more for commercial/industrial for audio file production and mastering, and was less suited to ripping and consumer tracks and in the early days wav meta data tended not to be supported by consumer software… although wav later unofficially supported ID3 tags as well.
As far as what sounds best… we’ll theoretically they should identical, but different systems are going to decode them in their own way and that may lead to a preference of one format over other due to side effects such as very low level digital noise caused by the decoding software execution.
Both are relatively straightforward to decode (and incidentally the decode effort for FLAC is constant irrespective of the compression level assuming it is compressed). However decoding wav into PCM is more straightforward than decoding FLAC to PCM and typically requires less processor processing time per sample.
As far as ‘degradation’ between local or remote FLAC files, I think it’s fair to say there is none. However the streaming distribution master may well be different to your local CD rip master, especially if that master is also used to create AAC, Ogg, or MP3 … perhaps in terms of average audio density in mastering. (LUFS). CD can often be mastered to also have lesser dynamic range than say a master destined for a streamer or lossy transcoding…. So CD can sometimes sound more in your face than the streamed equivalent, as the CD master is more compressed. (Has a smaller LUFS number).
Some say and I agree, since lossy formats and streamers have become prevalent over CD, dynamic range has started to return to popular music. A very compressed and normalised track will often sound very poor when transcoded to MP3 or AAC. Spotify for example has a standard of -14 LUFS and apparently enforces it which imposes quite a headroom for non normalised peaks… which can be good for dynamic range.
But if the FLAC is sent to the streamer without being transcoded by a UPnP server or similar, it’s the streamer that decompresses it eventually. You can’t feed FLAC as such into a DAC as you know, so in that sense it’s automatic, in the context that was discussed. The DAC always sees the PCM, whichever way it was originally packaged.
Yes assuming your DAC doesn’t have an Ethernet interface … (what some call a streamer DAC) then yes its interface will be SPDIF encoded format or a USB audio format.
So yes FLAC and WAV need to be decoded and the extracted PCM transcoded to SPDIF or USB audio format before being sent to the non network DAC
I think some people, and not saying you do, think SPDIF or usb sends pure PCM… it doesn’t it’s encoded PCM … just like wav or FLAC encoded the PCM, just a different format. SPDIF and USB also adds relative time into the PCM encoding as a data stream format.