advertisement


USB audio coming of age?

Status
Not open for further replies.
Server to streamer/computer is reliable TCP/IP FTP protocol

Streamer/Computer then transcodes to WAV (if necessary)

Then data into DAC via half duplex USB 'pipe' via PCM...preferably asynchronously into a buffer

Yup, that was my understanding.

So how can anyone claim that a .wav file transmitted along the USB bus is in anyway a "digital audio stream". It's just PCM data, surely?

Chris
 
But it strikes me that Beresford would have included async USB in his kit because the market demaded it. If it's a case of no async USB = no sales, you include async USB.
It is on his up and coming latest offering:eek:
 
Yup, that was my understanding.

So how can anyone claim that a .wav file transmitted along the USB bus is in anyway a "digital audio stream". It's just PCM data, surely?

Chris
Yes it's just data- and the meaning is arbitrary- after all if it were not you would not be able to send DSD over it. (incidentally I assume you can still send DSD over as you can with S/PDIF?)

I guess it's a stream in the sense that it is serial and that at least with audio "asynch" usb (but not proper bulk mode) there is some relation between the rate it flows and the rate it plays.
 
Yes it's just data- and the meaning is arbitrary- after all if it were not you would not be able to send DSD over it. (incidentally I assume you can still send DSD over as you can with S/PDIF?)

I guess it's a stream in the sense that it is serial and that at least with audio "asynch" usb (but not proper bulk mode) there is some relation between the rate it flows and the rate it plays.

What I find difficult to understand is why anybody bothers with any other transfer protocol other than TCP/IP.

I have put this question to the forum on a few occasions, and never received an answer, let alone a satisfactory one.

Chris
 
So how can anyone claim that a .wav file transmitted along the USB bus is in anyway a "digital audio stream". It's just PCM data, surely?

Chris

I have an opportunity to correct my misunderstanding. Functionally, I see USB into a dac the same as SPDIF into a dac, especially if my SPDIF was served up by a USB/SPDIF convertor.

So is it a (.wav) file that is sent then reconstituted, or is it a data stream decoded on the fly ?
 
The Beresford DAC in question isn't equipped with async USB. I found the AoS thread about it. I don't find him an authority that inspires confidence. Others miles may vary.

Paul

I have serious doubts about his actual input into the designs. Weren't they originally just OEM units with new front panels?
 
I have an opportunity to correct my misunderstanding. Functionally, I see USB into a dac the same as SPDIF into a dac, especially if my SPDIF was served up by a USB/SPDIF convertor.

So is it a (.wav) file that is sent then reconstituted, or is it a data stream decoded on the fly ?

Decoded on the fly, like on S/PDIF, but buffered (sent in packets) unlike S/PDIF.

Another important difference is async vs. isochronous. This relates to the synchronisation of the transfer rate. isochronous works just like S/PDIF in that packets are "pushed" down to the DAC at a rate determined by the computer. If the DAC uses a clock that is synchronized (using a PLL) to the incoming data clock, everything is nice and dandy, but there is at least a theoretical possibility that jitter on the computer-originated data clock can affect the audio clock. If you want to totally isolate the incoming data clock and the audio clock, you need an asynchronous sample rate converter to "bridge" the two clock domains.

Async USB has a signalling mechanism by which the DAC can regulate the delivery of the data, so no ASRC is needed.
 
What I find difficult to understand is why anybody bothers with any other transfer protocol other than TCP/IP.

I have put this question to the forum on a few occasions, and never received an answer, let alone a satisfactory one.

Chris
It's a good question. I suspect that many/most audio companies just don;t have the expertise, plus there's the issue of software flexibility (will it work with everyone's favourite music player.) I know that Linn do it that way but I was under the impression that it was a bit limited until someone worked out the LMS hack - or am I mistaken as I never had one I didn't look into it too closely.

I bought my ps audio dac with onboard bridge to take advantage of Tc/IP input, but they just couldn't get the software to work, so I ended up just using my SBT. Shame.

However I rather hope that the latest wave of RPI (or similar) playing software and I2s output might perhaps be turned into a sort of universal dac streaming input which will enable dac manufacturers to give their dacs in effect a TC/IP input. If the mainstream audio dac manufacturers don;t go for this then they may lose out because there are plenty of cheap i2s dacs boards appearing to take advantage and it seems that pre-built models are on the way.
 
Ok, thanks for that. Still seems to me that using a network connection (TCP/IP via wifi or RJ45) is not all analogous with transfer over usb. I viusalize the streamer thingy (Linn whatsit, Sqeezebox, etc) as reconstiting a file of audio data, which is then decoded into a digital output (PCM), first, and then decoded to analogue, either internally or externally.

Usb audio, for want of a better term, is sending the PCM stream.

Am I off track ?
 
I have an opportunity to correct my misunderstanding. Functionally, I see USB into a dac the same as SPDIF into a dac, especially if my SPDIF was served up by a USB/SPDIF convertor.

So is it a (.wav) file that is sent then reconstituted, or is it a data stream decoded on the fly ?
What is there to 'reconstitute' in a wav file?

In both cases, USB and TCP there is a logical channel between sender and receiver into which data is put at one end and retrieved at the other. What goes in first comes out first. On the wire the data is transmitted in bursts. The USB standard defines various types of 'end point' so a USB audio device can return track changing commands to the source, while continuing to play.

In both cases the clock that is used to transmit the data is completely independent of the audio sample rate clock.

In general for USB the data that is sent is direct representations of the individual PCM samples, there are agreed formats for 16, 24 and 32 bit values. For TCP it's possible to send compressed data for the destination to unpack, the protocols for this have to be streamable, so that the decoder can pick up the stream anywhere in time and quickly synchronise to it. This allows Internet radio, or live TV streaming, to work.

SPDIF is somewhat different in that the data is absolutely real time and arrives at a constant rate rather than in bursts.

Paul
 
Ok, thanks for that. Still seems to me that using a network connection (TCP/IP via wifi or RJ45) is not all analogous with transfer over usb. I viusalize the streamer thingy (Linn whatsit, Sqeezebox, etc) as reconstiting a file of audio data, which is then decoded into a digital output (PCM), first, and then decoded to analogue, either internally or externally.

Part of the problem is that we still don't have an unambiguous definition of a "streamer". Let's take the squeezebox as an example - it has 2 components, the server (that handles files, either from local storage or from network-attached storage) and sends the data to player units over the network. That data, in the format of a FLAC, WAV or mp3 stream, consists of a file header and the actual data content - in the case of WAV, pretty much the raw PCM data, in the case of FLAC the PCM data losslessly compressed, and in the case of mp3, the lossy perceptual encoding of the data. At the receiving end, the player pretty much just hands the data part of a WAV file to the DAC, unpacks the FLAC encoding "on the go", or decodes the mp3 stream, and passes the stream over to the DAC.

Your typical DAC usually only understands pure PCM, so can not decode FLAC or mp3, but in the case of a WAV file there is not a huge difference.
 
Into my Young dac I run my pc with jriver USB out into a ifi iusb then into Young USB input.

The pc has the music files on internal hd but can also load the same files that are on the nas in another room.

Also into the Young is my Cyrus StreamX feeding the signal from wired Ethernet from the nas and into the Young using Chord Signature coax.

My Cyrus cdxtse cd transport feeds the Young via optical cable and my Sony Blu-ray player goes into the Young via bnc.

To my ears the USB from pc sounds the best and a very close second is the cyrus streamer via coax and the CD comes 3rd and this was also how I hear it using my M-dac.

The differences are very small ill grant you but they are there with my Young dac my m-dac and my Cyrus dacX.

To my ears using the USB out from the SBT using EDO made a nice upgrade over its coax and optical.
 
Peter, why do you think bits aren't bits? If they aren't, what are they, and how then do computers work?
 
Status
Not open for further replies.


advertisement


Back
Top