Ok I know I’m going to annoy a lot of people with this thread but in the pursuit of knowledge I shall proceed. So as some of you may know I’m a huge SW1X (pronounced S1X) fan. Possibly as big of a fan as I am of Naim.
Slawa (founder and designer of SW1X) has repeatedly told me about ensuring the music playback is “local” and to avoid streaming when possible. He is adamant that streaming is a convenient but compromised solution to music playback. Well for the longest time I pursued with streaming via Audirvāna on my MacBook to my Abbas DDC. I’ve recently bought two sets of speakers of which the SW1X Monitor 963 Field Coils especially are super revealing, the other being the legendary Naim DBL speakers. I only mention this as perhaps the reason I’m able to reveal the following is because of the uber transparent window into the audio that I’m now getting with these speakers.
So late one night I don’t know what came over me but I downloaded a free trial of a software that downloads streaming files just to put some local files on my SSD to put this claim to the test. Well I don’t know why I ever doubted the guy because YES the local file sounded more natural, less edgy than the same FLAC file being streamed through Audirvāna.
And now onto tonight where I decided to purchase a full version of the software to download some more files onto my drive as I edge closer to the homecoming of the NAC S1. This time, I decided to convert the FLAC to WAV as I had this feeling that surely the uncompressed WAV file would do even better. Well I’m happy and slightly annoyed to report my suspicion was spot on. WAV sounded better still. The annoying part is WAV takes up more space and I’ll need to figure out a way to store all my 5000 tracks onto a solid state drive when the largest one I can find is 8TB.
My reason for reporting this here is that I ran a search and found many posts about converting WAV to FLAC but none to do it the other way around. Well if your system is as good as you’d like to believe, I’m here to tell you - feed it the best file which is a stored WAV file NOT FLAC
P.s. my tube preamp and tube monoblocks are sounding so good of late that I’m somewhat regretting my purchase of the NAC S1. I’m not entirely convinced it will be better. I keep saying to myself, surely it must be.
In my system WAV files have always sounded better than the same music on FLAC, and local streaming always outperforms internet streaming for me, so I am not surprised by your findings.
It’s highly dependent on the effect the processing of the FLAC to unpack the data causes. Both FLAC and WAV will contain identical data if the FLAC is created from the WAV, but any processing workload is not free in terms of noise generation.
To what degree that is mitigated by a streamer is another matter entirely. You can of course offload the unpacking of FLAC to WAV at the UPnP end such as in Asset. Or take it further and have both unpacked to LPCM if the streamer supports that over the network (it should as it’s basically just the payload portion of the WAV file).
On Gen1 Naim streamers, Naim themselves were fairly upfront about the fact that those devices were specifically noise optimized to cope best with WAV.
I must admit though, I find the difference to be somewhat exaggerated. To me it’s another case of yes there is an audible difference but an obvious or big difference is not something I can agree with. I’d describe it as very subtle.
I agree. Although I like AIFF which is the Apple version of uncompressed music files. Which was designed from the ground up for music and metadata storage.
That’s likely going to depend more on whether the transport doing the unpacking is big or little endian at the processor level. If it’s natively little endian, WAV will incur less work than AIFF. If big endian, the reverse will be true.
So depending on the streamer, I imagine there could be some variability here. Though dealing with sample endianness is likely less workload than decompression of an archive.
Many years back I recall planning my NAS storage strategy and testing both formats as carefully as I could. I felt that there was a subtle difference that mostly favoured WAV but not on every track or genre. I set up my Asset UpNp to convert flac to wav as it plays them, keeping my flac files for the benefit of better editing of metadata and more compact files. Seemed the best compromise. Many people do this.
On the old forum when the first streamers were arriving this was debated often and I think the previous CEO of Naim whose name I now forget (Stephenson?) made a rare contribution to confirm that early Naim streamers sounded better with wav.
So this debate isn’t new (or annoying). Lots of people have come to a similar conclusion but generally found the difference very subtle.
Finally, by comparing online vs local streaming (as I think you were doing) you are adding a very significant variable. Only comparing identically located local files stored in each format can you really reach a conclusion. Local vs online is a much bigger difference to flac vs wav in my view.
Bruce
PS shouldn’t this be in the Streaming section of the Forum? Just an idea.
Just a friendly rundown of the differences between WAV, FLAC, and PCM:
WAV is basically raw PCM, while FLAC is like a zipped-up version of PCM, i.e. compressed but still lossless! after unzipping or decompressing, a FLAC file and its WAV twin are identical down to every bit; the PCM samples match perfectly. Any differences you might hear usually come from how your gear handles things—like decompression noise, CPU/memory use, or how your streamer’s firmware works with each format.
The original streamers are better playing WAV, but I understand the 2nd gen and later there is no or little difference.
So my take on that it might apply to other brands as well …. only experiment and personal preference can tell.
I used to be WAV, then Asset for Synology arrived and I converted to FLAC and transcode (“play as”) WAV ….. I can easily “play as” FLAC and that confirms the streamer does sound slightly better playing WAV.
Some hear that a straight WAV rip (with no metadata or FLAC conversion) sounds better than a WAV file that has been converted or has metadata. However, once metadata has been added or the file has gone through any conversion to FLAC and back again to WAV, any SQ advantage is lessened - I recall reading a fascinating article about this some years ago (by whom or from where I can no longer recall though - was it Hifi Plush or Critic?). I did try it out on a couple of music pieces and do think there’s something in it though, with a slight favouring of the straight WAV file. It’s what persuaded me to do multiple rips, so I have my CDs in straight WAV (ripped via the Core or HDX), FLAC (ripped via DBpoweramp - ideal for non-Apple portable devices) and then ALAC (ripped via iTunes/Apple Music - ideal for iPods).
These are the downloads available on Bandcamp and I usually download AIFF and WAV only. I used to include FLAC but stopped using it after a while. Unfortunately I couldn’t hear much difference between the formats. It’s only when I compare WAV/AIFF to MP3-320 I hear a difference, MP3-320 is more shouty with the bass less developed, WAV and AIFF smoother and less grain, better quality bass and overall higher quality sound.
There is a huge difference in the size with this particular track, and MP3-320 is significantly smaller hence the quality is not very good.
You can create FLAC files with no compression (dBpoweramp supports this) so you effectively avoid the work of decompression in the streamer, but get metadata…
Edit: I might also add you can vary the amount of compression too, less compression, less effort to uncompress. I did notice somewhere that Melco FLACs were larger files (less compression) than std. dBpoweramp ones, so there could be a sweet spot between full uncompressed and saving some space for larger collections.
It depends on what component is unpacking the FLAC. Most people have NAS and it’s the NAS that unpacks and transcodes.
With a modern NAS this is a very lightweight task and causes no issues.
IIRC has been noted in the past that flac and wav sounding different or the same depends on the rendering device, I guess data handling capacity.. out of curiosity how much RAM is in your Macbook, and do you run it optimised with dedicated USB bus and nothing else running on the computer whilst playing music? (I use a Mac mini with 16GB RAM, and which has the advantage of also being able to be run headless, with no screen).
Your comment on local vs online streaming fits my expectation but not verified as I have never subscribed to any streaming services. However others have reported no difference, and I suspect that things like the qyality of internet connection, the local access point, and the internet traffic. Also if your SSD is within or directly connected to the Macbook your music playing doesn’t invove music going over a network, obviating the potential RF noise related effects music played over ethernet may have, unlike if you stream online, so that may be contributory. Others more commonly playing music across a network may experience differently.
Regarding your speakers, as I observed in the thread on field coils, the fact that it is a field coil will not be what’s making them revealing – they would sound exactly the same with the same magnetic gap dimensions energised to the same field strength by permanent magnets. It will be the remainder of the design of the driver and any enclosure behind it that make those speakers more revealing from others you’ve compared with, though crossover may also play a part.
I have to say that reading through this thread makes me even more glad that I abandoned streaming and returned to CD replay. I didn’t stream from on line libraries, or even across a LAN, but rather direct from a Melco music library into a DAC. Some potential issues were side-stepped because of this but many remained and it was clear to me that to deal with these effectively to my satisfaction would require significant expenditure which I just couldn’t afford.
Some of the issues I encountered may have been specific to my own set up but I’m sure some were more widely applicable. CD just gets rid of them all.
Does it? Assuming that the NAS is what runs your UPnP server as is usually, but not always the case, my understanding is that whatever format you choose would be served to the streamer as WAV, FLAC or whatever, and the processing done there.
In the case of files being transcoded ‘on the fly’ if you choose to do this, your FLAC file would still be sent to the streamer as WAV. The only UPnP servers I’m aware of that can do this are Asset, Minimserver and the 1st gen Naim servers (Unitiserve/HDX/NS01.)
Having had no issues whatsoever with local streaming, and a sound quality that at least matched CD play, I haven’t understood what you found problematic, especially as your setup also avoided the potential noise issues of playing over a network. Maybe it was specific to the gear you used. (I’ve only played via a Melco at a dealer, otherwise my streaming source and DACs have been different from yours, with never any problems through my journey from initial streaming setup to final: First cheap basic NAS to ND5XS, then same with Hugo DAC, then Mac Mini running Serviio as NAS into ND5XS into Hugo, then Mac Mini running Audirvana through Gustard USB-SPDIF converter into Hugo, and finally Mac Mini running Audirvana into Dave,)
Hi Chris, yes you’re right ….. In too much hurry again, I was meaning to qualify the transcoding process.
As I understand it, transcoding unpacks FLAC and converts to WAV before sending to the streamer.
In all cases, all streamers finally convert whatever they receive to PCM.
Obviously FLAC and WAV contain the same audio data, however your renderer may create more or a different noise profile running the WAV expansion code vs the more involved, though not demanding, code for FLAC expansion. Because of this you may notice a sound rendering difference between the two on your system. But it WILL be system dependent.
I have set my system up and used components where they are identical in terms of rendering. It didn’t used to be the case, with my first all Naim streaming setup 15 years ago WAV sounded better.
I separate my DAC from my streamer which I believe goes a long way to getting equivalence.
FLAC offers more utility and saves space of redundant packing data packing space compared to WAV.
In my experience of streaming over the last two decades plus, I have found if you can hear a difference between FLAC and WAV on a component, you are more likely to hear differences from underlying network operation and cloud streaming. I have found it better to strive to isolate such artefacts. It helps prevent obsessing and get in the way of enjoying and appreciating your recordings.
Remember for cloud streaming, the streamer does have to work a little harder than local DLNA streaming as it needs to decrypt the session data. Local streaming with UPnP doesn’t encrypt.
