24bit .wav files not recognised

>  But a bloke from Grahams said he preferred the sound of FLAC. 

Amazing .... are these magic FLACs he was referring to. 

When you extract music from FLAC it is the same as when you extract it from WAV. 

It is bit for bit identical you can prove this with a program like XLD on a Mac. 

I have read that the extraction process. which takes no more than 1% of extra processing power on my Mac G5 processor, but not sure what it does on Naim music player, can have an effect on the sound. If it does then I'm concerned that Naim haven't addressed this. 

However, for FLAC to sound better means the extra noise added by the extraction processes is beneficial; like saying an amp with a bit more distortion sounds better, but then some people do say this and use valve amplifiers from you know where. 

Personally, I've never been able to hear a difference, but my system is quite modest so perhaps it masks it. My crude measurements of capturing the data streams from various format and players lead me to believe that the output from these is the same and that the DAC see the same thing. 

The benefit of FLAC is the way it does tagging; the downside of FLAC is iTunes can't use it so if you use iTunes then it is a pain. Naim supports ALAC which solves this problem and again sounds the same to me. AIFF is the iTunes friendly version of WAV and is uncompressed and supports Tags. [AIFF comes from the Commodore Amiga days - it was the Electronic Arts interchange file format]. 

I don't use WAV because I've never cracked its tagging idiosyncrasies.

If you don't need iTunes compatibility then I'd go for FLAC.   

Guy, How on earth can Naim address how FLAC works?

I appreciate you don't hear a difference between FLAC and WAV. I can, on a Naim Uniti, which isn't exactly Naim's top end. The notes at the higher end of the spectrum are slightly woolly/fragmented/less stable around the edges. Yes I've done it blind, no I don't get it right every time as it's only a tiny difference, and if I hadn't heard the WAV I'm sure I'd happily listen to FLAC. But any processing that isn't necessary might as well be avoided and as dbpoweramp and asset do a perectly good job with my tags I see no benefit at all in a compressed format. (Aside: high end slr camera users shoot in RAW rather than jpeg to avoid that processing and compression, the same principle)

XLD is only good to count up 1s and 0s, not interpret the subsequent sound that comes out.

1101 011 is not the same as 1101011, though quickly read it would appear to be.

Believe it or not computers, even of the Apple variety, do not read from memory with 100% accuracy. The 'reader' can go back and check if it has missed a digit but in music terms (and video) it's too late. The rest of the 'packet' has moved on and no buffer can keep it held up while it's accurately reinstalled. With a spreadsheet, a web page etc etc, yes it can be slotted back in and restore the faithful reproduction. Overly simplistic but hopefully you can get my point.

As for the Graham's chap. Maybe his ears just prefer that sound. I can't imagine  finding a speaker better than a Wilson Benesch Square One for the money, but plenty of folk do. I've even heard that some people like paying extra for a particular brand of computer that reaaly only does the same as every other brand. It's a grand world

Andy - RAW to JPEG is throwing away information due to compression. You can't reconvert the JPEG back to RAW and recover that information - its a lossy process. 

WAV vs Flac  - The issue isn't the recovery of the original data but the extra noise generated by the processor within the streamer ( conducted and radiated ) when doing this process compared to playing an uncompressed file.

James

> Guy, How on earth can Naim address how FLAC works?

Andy - they don't really work with FLAC, but the PCM inside it - having a sturdy envelope rather than a flimsy one should not change the content, though it is easier to open the flimsy one. 

Simply build a player that does the decompression without a drop in sound quality - Naim have said if you stream PCM from a Mac Mini then the fact it started life as FLAC, WAV, AIFF or ALAC makes no difference because they only see it a low jitter stream of PCM. 

My Apple computer output identical PCM from these formats in the samples I captured - it didn't miss a digit. If it did then may of the claims about the advantages of ripping would be void. 

> XLD is only good to count up 1s and 0s, not interpret the subsequent sound that comes out.

1101 011 is not the same as 1101011, though quickly read it would appear to be.

Yes, but the stream will not have space in it - there will be jitter, but it is very small - added to which the Naim DAC buffers. I can't measure it, but do we really think the jitter inside the Naim Uniti is so bad there are audible differences. Again you hear what you hear: I'm not questioning that, but I think the explanation that it is down to missing bits or internal jitter is unlikely.   

My understanding was Naim doesn't use the computing power of a Mac to reduce noise, so the difference in processor use would be much greater, but I've not checked this a fact. I can accept that this noise may be audible, but surely there is a way around that that Naim can engineer, 

Meridian and Linn engineer their players so even when on-board they do not impact sound quality from using FLAC. 

My hope is Naim will achieve this with the NDS. I don't want to keep my files as WAV because of the way it treats metadata - if Naim could do it with AIFF that would work for me. 

The sound from a WB Square One is measurably different from that of another speaker such as my TD510s ... so agree that we should hear differences easily - loudspeakers are the easiest things to tell apart. 

However with identical streams of PCM - they are the same - it is not dropping digits that can make them sound different, but some form of noise being introduced, which is hope is what Naim has banished with the optical isolation in the NDS. 

The system that I used to capture the PCM is through a Naim UQ digital out which I can send back to my computer - in terms of bits, what goes in comes out - well at least for the files I've tried. 

All the best, Guy 

+1 - The RAW-JPEG analogy is like WAV-MP3, not WAV-FLAC. 

WAV Vs Flac: Yes, but in my experience processor load makes little, if any difference to SQ. I can be listening to music (ALAC via iTunes to Mac Mini internal sound output device to SPDIF to SuperNait) whilst I simultaneously rip a DVD and simultaneously process another video file to play on my iPad, thus using 100% of the processor power; and the music sounds just the same. It is hard to imagine that a trifling couple of extra processor % to decode a FLAC rather than a WAV would be important. 

Guy, I don't know if you are being very conceptual in your description. The streams you are referring to are these SPDIF? If so SPDIF doesn't send a PCM streams as such, the SPDIF protocol chops it up somewhat. However the timing of the SPDIF channel frames is used to recalculate the sample clock using an established conversion formula. However if the timing of the frames is slightly inconsistent (transport jitter) the calculated sample clock will be varying, unless buffered and either locked or phase lock looped.

Winky - I was talking about processor loading within the streamer itself when dealing with compressed files. I think Naim mention this themselves in the white paper on the NDS.

James

Winkyincanada, if you were using a general purpose CPU in a multitasking environment there will almost certainly be no difference in the 'noise' associated with decoding a WAV to a FLAC.... 

However if you are using an optmsed micro controller in a an environment of noise sensitive digital and analogue electronics, then the difference in decoding WAV and FLAC becomes more relevant unless the decoder is isolated to the point of any induced noise is at or below the digital clock or analogue noise floor.

A good analogy of FLAC is a ZIP file.

Simon

You mean this bit?

"Uncompressed audio data will always give better sonic
results than compressed. Even lossless compression
may not reproduce audio with equivalent quality to the
uncompressed original as the processing required to
uncompress the data increases the computational load.
This raises the power supply noise floor, which impacts
on the sound quality."

If anything is likely to generate noise, it is the distinctly cheap-and-cheerful non-audiophile SMPS in my Mini. But yet it doesn't, even when the Mini is working MUCH harder than the processor in a streamer ever would. I know it is a separate box, but connected through the mains to my HiCap and SuperNait.

I varied the processor load by about 2500%. From 4% to 100% and I can't hear any difference. Cloth ears, I guess.

Increased noise on the power rails within the streamer - see Simons post above. 

Winky, sorry you might have misunderstood me, a multitasking CPU such as in a MAC or PC will be generating lots of noise irrespective of whether a WAV or FLAC is being decoded, so the noise produced will be  far greater than the discernible difference between decoding a WAV or FLAC (even at 4%, which doesnt tell you much anway in an OS envionment in terms of CPU cycle noise). As you are not coupling your DAC to the motherboard powerlines or siting the DAC in close proximity to your motherboard this doesn't really matter anyway, and if you did the DAC would almost certainly sound mediocre.

Microcontrollers do produce a noise signature when executing a specific algorithm, I remember in a previous job I was looking at smart card reverse engineering in a lab in Israel, and by applying signals to the inputs the 'code' could be reverse engineered by sensing the EMI produced... Fascinating it was..... and that EMI was a miniscule fraction of the EMI that a Naim streaming board micro controller would produce.

Yes absolutely right Simon ... sorry I wrote it in a confusing way. 

Really just trying to say my Mac Mini extracts the same data irrespective of the container and shoves it down the S/PDIF - no way from the receiver at the other end of the S/PDIF to know how it was stored. 

I'm guessing this would true of the new HDX and US in to a Naim DAC ... they would just send out the data through S/PDIF and the DAC would be none the wiser. 

So I'm hoping that within the NDS that the way it is extracts the bits in its renderer and ships them to the DAC is such that it makes no matter what format the file was in ... even though the NDS will know because I'm sending them over Ethernet/TCP-IP container and all. I'm also hoping the optical construction and isolation will make any extra noise of no consequence. 

This means no messing about from my point of view .... if I download a FLAC then all I have to do is play it. if it is ALAC or AIFF then likewise. I can be very lazy and the NDS will do all the work. 

As I understand it the Vortexbox is unlikely to offer a transcoding option and Asset is a non-starter for me unless Spoon does a Linux or OS X port.

I wonder does the new 2TB UnitiServe transcode before sending to a Naim streamer, if it did then it is getting ever more attractive, as I can easily put my top 2,000 albums on it without any re-ripping  ...  I would still need to able to update it easily though.  

All the best, Guy 

Right, back to the drawing board then. Oh my, Amy Macdonald is caressing my ears (No dear, I was talking figuratively..... Yes dear, shutting up now). What were we on about?

Hi Guy.. I suspect you are right, the big benefit of the NDS might be its A1 out of the box, where as with NDX->NDAC/555PS takes some tweaking and optimising for it to sound A1, well that's for Redbook, I am hoping the NDS improves on Hidef which to my ears is often not as enjoyable as Redbook  on my system (if you exclude the sonic fireworks  such as on Meet Me in London)