How to do USB audio

Update to the above. All being well I should be getting a 192k/24 USB ADC tomorrow. Be interesting to see if I can record to the NAS using it! :-)

Jim

Now have the Benchmark ADC1 USB to test. Worked right away with the RO audio modules. :-) Did a 96k/24 bit recording and the results sound OK. But I’ll need to do more careful tests on quality to say more. Unlike the other ADCs I’ve tried, it uses 3 bytes per 24bit sample transferred, not 4. Alas, it seems it can only output 192k/24 from its spdif/AES sockets. USB is limited to 96k/24. But it should be better quality than the previous ADCs I’ve tested…

Jim

I now also have a NuWave ‘Phono Converter’ to test. This a fairly ‘high end’ device for those really keen on hifi. Quick tests show it works OK with my ARMiniX as an audio ADC. First time I’ve made a 192k/24bit recording using a USB ADC! :-) It actually lists even higher rates, but at present my recorder program (and my analysis programs!) can’t cope with > 192k!

Jim

I’m testing various DAC devices and so far it looks like that most of them are working ( have tested a few player-only DAC devices such as the dragonfly, meridan and so on)
By far the most natural sounding DAC was a so called NOS DAC ( non oversampling) the NOS DAC octave MK2
www.nosminidac.nl
For the best listening result you need a good headphone or a decent hi-fi set ( for the record I,m playing my files on a
Quad 44. 405-2 amplifier wit a couple of QUAD 988 ESL speakers)

All we need is a ‘good’ player program for RO – ( FLAC support and/ or playlist management)
I’m willing to pay 100 uk pounds to the one who will write such a program :). )

You can use one of the programs from my software page to generate files that produce much the same effect as NOS if you wish. So generate the effect on other DACs. They just need to be able to play at x2 the rate of the orginal file. :-)

If interested, have a look at

http://www.audiomisc.co.uk/software/ARMiniX/Upsampling.html

The WAVupsampler program will create various effective digital filter behaviours, inc ‘NOS’.

Good luck with the Flac player for USB! Something I guess many will want. I’ve not attempted it because I fear it is above my level in programming skills! I keep hoping that someone will ‘upgrade’ DCD or PlaySound and their relevant modules to do this. But not something I can do myself, alas.

Jim

I think it needs someone to enhance SharedSound to handle USB devices in addition to any built-in audio interfaces. There would have to be a means to enumerate available devices and to choose from them. Logically, SharedSound should call the USBAudio module.

Once that is done, all the existing audio applications can use USB audio devices.

I’m not volunteering for the job; I will be spending all my programming time for the foreseeable future on Bluetooth.

Agree with the comment wrt SharedSound for general use. My only concern about that is if it fiddles with the samples in situations where the user doesn’t want it to! Personally, I want the LPCM sample values defined in the file to be the ones reaching the DAC. So this comes down to the details of how any ‘sharing’ or ‘volume’, ‘mixing’, etc, is done. Raises the old questions wrt blocking versus mixing, etc.

However using SharedSound makes sense in general provided direct access to the USBAudio remains available as an alternative for those who want to avoid mixing, etc.

FWIW At present the RO USBAudio route is about the best on offer for serious audio, precisely because it is relatively direct. I’ve been working on a Linux version of my USBRecorder. It works, but getting ALSA to co-operate and shove other layers of mixers, pulseaudio, etc, etc, out of the expletive way can be far harder than it should be! :-/

Jim

I agree with Dave, I think we need some easy way for general sound to be output via USB, whether that’s playing MP3s, or even the system beep! Most people will probably be happy with some sort of sound coming out of their USB speakers, even if it’s not absolutely perfect in every way! For audiophiles such as Jim where even the live performance isn’t high enough quality, then the current specialist apps can be used for playback.

Personally, I want the LPCM sample values defined in the file to be the ones reaching the DAC.

Even you don’t always, Jim, and the wider world often won’t. The most basic example is on replay where the samples in the file are of a different word length than the DAC. If you don’t change the samples, then you don’t play the file at all. Surely some audio is better than none at all.

It would be sensible to have some settings available for how to change word lengths.

I think it needs someone to enhance SharedSound to handle USB devices

I don’t see why SharedSound is implicated, it doesn’t know about any of the existing hardware so why should it deal with USB any differently?

Most of RISC OS is built as a software “stack”, with hardware at the bottom of the pile and higher level concepts higher up. In the sound system it’s SoundDMA (one of its many variants) that sends the output buffer to the hardware, every module above that just knows about samples, so I would imagine a replacement SoundDMA is “all” that is needed to shovel data out. You then get the 8 bit and 16 bit sound system support for free.

Even you don’t always, Jim, and the wider world often won’t. The most basic example is on replay where the samples in the file are of a different word length than the DAC. If you don’t change the samples, then you don’t play the file at all. Surely some audio is better than none at all.

I think I covered that in my “makes sense in general…” before the provided.

Yes, you do sometimes need to trim or pad zero bytes. And its fine if SharedSound can do that without otherwise tampering if the user only wants the trimming or adding of padding.

Alas, the history of RO systems is littered with examples of flawed sound systems. So I’d be wary of pinning all on having a new SS work perfectly as a gateway to USBAudio.

At present we can bypass that and get results that match or surpass any other OS. If you’re going to make ‘via SS’ the default route the challenge would be to not foul up access to that level of quality.

However as I said, provided we can still get ‘direct to USBAudio’ as now, that is OK because we have a choice. People can then decide if they want quality or convenience if SS can’t simply pass on the info without unwanted changes.

FWIW I agree with Kees that at present the main problem in practice is the lack of ability to play Flac. I’d be happy to add other features like jumping to a user-chosen point, pausing, playlists, to my simple player. But I’m afraid flac is probably beyond me.

Bear in mind also that I’m already looking at USB devices that do DSD/DoP and rates above 192k, etc. DoP has to be send bit-perfect or risks being totally scrambled. I’m not a fan of DSD, but for all I know, it will become popular. Avoiding needless alterations also avoids problems in future when ‘new’ situations arise. And I hate to think what an LPCM SS ‘mixer’ would do to DoP! 8-]

Jim

Just to let people know I’ve now put up a webpage on the results of my tests on two high-quality (and high price!) USB ADCs. The Benchmark ADC1 and the P.S. Audio NuWave Phono Convertor. You can find them at

http://www.audiomisc.co.uk/HFN/ADC/HighQualityADCs.html

I ended up buying the ADC1 I was sent to test!

Jim

I finally bought a pi (a b+) and have USB Audio working on it. It plays fine with a cheap usb1 44/16 dongle via usb hard disc, sdfs and sharefs but the sound breaks up with my USB2 device. There’s still a few problems but there always is :-)

I finally bought a pi (a b+) and have USB Audio working on it.

Excellent! (having snipped the problems 8-] ).

I keep meaning to buy an RPi, but get distracted by other things. FWIW I’m currently still working on the NuWave and DSD. Hoping to be able to record, decode and play DSD since the DoP transfer method lets it ‘pretend’ to be LPCM.

Jim