Spot on Chris. @Thomas During a visit to Signals, the Naim dealer near Ipswich, we compared FLAC level 5, FALC level 0 and AIFF rips of the same source cd. The AIFF file size was 30% larger than the FLAC level 0 size of the same tracks.
For reasons that are beyond me, we also hear a consistent difference between FLAC 5, FLAC 0 and AIFF versions of the same cds stored on our Prestige 3 server, so we only use AIFF now.
We have considered using WAV but stay with AIFF as so many apple mobile devices use this music library and AIFF works best for us.
I would certainly rip to FLAC. It is lossless, that is, no drop in quality and significantly smaller than WAV. Also it is open source, hence not tied to any particular manufacturer and widely recognised. Even MS Windows players will play FLAC files.
Hi, I would recommend FLAC for storage over WAV, because less space is taken and not all consumer software reads WAV meta data. They are both lossless non proprietary PCM file formats. They can both store meta data, but as I mentioned previously some consumer software is not able to read WAV meta data… Naim is an example of this. Strictly speaking the method WAV uses to store album art is non standard, but has been defacto adopted. WAV can store more mastering/production meta data types, where as FLAC is limited to simply consumer audio track replay information. Most consumer rippers RAAT write WAV metadata simply store standard audio replay track info in WAV anyway.
On playback they both provide the same audio information, however more processing is required to read FLAC (BTW the processing effort to read is the same irrespective of the compression level), and because of this extra processing compared to WAV more digital noise can be created by the reading/parsing processor and detract from the audio experience on some (but certainly not all) devices.
So to get around this for those who are using equipment that does suffer this, then transcode is sometimes used in UPnP media servers, such as the FLAC is converted to wav for the streamer on playback. Some products like Roon do this automatically anyway.
With my 272 I preferred WAV to FLAC, but could not distinguish between straight WAV and FLAC transcoded to WAV on the fly. I think this is a fairly common experience.
On my Atom (and AURALiC streamer) I struggle to hear a difference between WAV and FLAC, so I use FLAC, primarily for convenience.
You should not rip to WAV with a Naim ripper. If you do so, the metadata of your rips will be saved in a proprietary database and you will be forced to use this database until the end of your days or, alternatively, re-rip all your CDs.
The question of which file format sounds best on a given system has no bearing on your choice: you can always transcode FLAC to WAV or to AIFF (or the other way round, of course) on your server and send to your renderer the format that sounds best on that renderer.
How would this translate to the extra noise, interrupt activity and increased power supply load introduced by reading a larger WAV file compared to a smaller FLAC? Does that influence things?
And would by this logic also not a higher resolution source file such as 24/96 or 32/192 sound worse than a 16/44 file on the aforementioned streamers, because of the extra processing power needed?
Do we have any hard data on this topic that could substantiate the claims? Measurements or spectograms perhaps? The WAV/FLAC discussion is a recurring topic, and it has been mentioned that even Naim itself advises to pre-transcode FLAC to WAV on first generation streamers. If that is the case, perhaps it would be good to have a sticky somewhere with actual references and measurements rather than anecdotal evidence?
The only reference to a statement from Naim i can find online is second hand and dates back to around 2011:
Naim weighed into the debate recently after making some detailed measurements of their streamers. They concluded that there were measurable differences in terms of processor activity and rail noise, but that these were not significant enough to translate to audible effects on the output.
I don’t understand too why Flac needs more extra processing vs Wav, because the Flac is smaller.
Since the thread I am struggling to find differences between Flac and AIFF. Sometimes I feel a very little bit that AIFF can sound better, but I need really to be hard concentrated.
My hearing was for now very accurate and could differentiate easily some tiny changes. But here it’s very tiny for me.
With Nds/555dr/ Melco N1z2/ ER/ MCRU, the source is enough resolving.
FLAC in its compressed state is smaller, its uncompressed before streamed to the player as raw PCM. So as far as the player is concerned, FLAC WAC AIFF, its all the same size & PCM
Because to make it smaller you need to perform mathematical operations on the data, which you have to undo to restore it to the original state. Same as a zip file really.
I understand that @GraemeH prefers replaying a FLAC file vs replaying the correspondent WAV file on the NDX2. What is strange or interesting about that? Even though the two files contain the same data, they are different and have to be processed differently by the NDX2.
The good thing is that every decent UPnP server supports transcoding. Thus, no matter whether one has stored the data in FLAC, AIFF or WAV, one can send them to a streamer in whatever format one finds that sounds best on that streamer.
I can do a near immediate change between FLAC & WAV. I find that FLAC resolves complex richly layered music more convincingly. Each distinct part slightly better articulated.
It’s slight, but noticeable with a keen listen - to my ears.
