My understanding is that as long as you, or the player software you are using, use ALSA services all is well. I use Squeezelite under Debian and the drivers miero has written work fine.
The reason it works with Squeezelite is that it supports streaming native DSD to it from LMS and it does not convert to DoP on output unless you add the -D option. This may not be the case for other players, such as MPD, that either convert to PCM by default or converts to DoP when enabled even though it uses ALSA. Where can I find Miero's drivers so I can test things here?
I wasn't talking about DSD, just straight PCM. Virtually all I play are CD image FLAC files.
Drivers are here: http://bbb.ieero.com/
Drivers are here: http://bbb.ieero.com/
I wasn't talking about DSD, just straight PCM. Virtually all I play are CD image FLAC files.
Drivers are here: http://bbb.ieero.com/
Sorry...I guess I have DSD go fever🙂
The driver with DSD suport is not-available yet. Currently it is not usable for other means than testing, because BBB does not have 44.1kHz clock so the DSD can be played only 9% faster of its nominal speed.
so the DSD can be played only 9% faster of its nominal speed.
That might be ok with Dance Music after a few drinks 🙂
That might be ok with Dance Music after a few drinks 🙂
Be bloody funny to watch!
Sorry if this has been asked before but I haven't seen it mentioned.You won't benefit from the FIFO if you use the cape, as everything will already be clocked only to the audio master clocks. No reclocking required. 🙂
The only way to do synchronous reclocking, AFAIK, is to feed the audio clock signal back from the Cape to the BBB. As there are two audio clocks only one is enabled at any one time, depending on which speed family is being handled by the BBB. To decide which speed family the audio file belongs to requires that the beginning of the audio file is first read & the speed family ascertained. Can the BBB change audio clock (provided by the Cape) on the fly to accommodate this scenario or is there some other clever trick?
Yes, the BBB chooses the clock which will be used to play audio data. This clock goes as input for the BBB and optionally also to the DAC if it is configured to the synchronous mode.
Yes, the BBB chooses the clock which will be used to play audio data. This clock goes as input for the BBB and optionally also to the DAC if it is configured to the synchronous mode.
OK, thanks
The BBB allows for changing this input clock while powered up & operating - i.e on the fly?
As far as I understand how this would operate - The start of the audio file will have to be read to establish the speed family, the correct clock then enabled & its signal used as input to the BBB - all this has to happen while the audio file is still being read - so some buffering is required. Am I correct in my assumptions?
Last edited:
OK, thanks - any more details of how this operates as it seems tricky to achieve? Are my assumptions correct?Yes, on the fly.
Both clocks have OutputEnable pin which can be easily controlled by BBB. Setting this pin to low disables the clock.
Both clocks have OutputEnable pin which can be easily controlled by BBB. Setting this pin to low disables the clock.
Yes, that's simple - I don't mean that - I mean that the BBB will have to have started reading the audio file before it knows which of the two clocks to enable. This enabled clock then has to be used as input to the BBB. Changing clocks on the fly like this is not instantaneous & so some delay/wait period is involved (some parts of the system has to be shutdown & restarted with the new clock, perhaps?). While this is happening the audio file still has to be read in real-time or some buffer used which can deal with this wait period.
This is done deep in the kernel driver. The ALSA library before feeding data invokes snd_pcm_hw_params*() call that gets propagated into the kernel. Kernel will switch the clock and then returns. So when application will start playing the clock will be already switched...
Real time kernel?
Would real time kernel be an improvement?
https://groups.google.com/forum/#!category-topic/beagleboard/beaglebone-black/N7vSFg5FLDo
(I don't have the expertise to build an image yet. First I need to configure a computer with linux)
Would real time kernel be an improvement?
https://groups.google.com/forum/#!category-topic/beagleboard/beaglebone-black/N7vSFg5FLDo
(I don't have the expertise to build an image yet. First I need to configure a computer with linux)
If you need to control a nucular power plant, it is a good idea to use software certified for realtime operation. After all, it would be slightly annoying if reactor safety scram was slightly delayed because Windows Update decided to install a "critical update" during a tsunami.
In the case at hand ... mmmmmh, it would probably be useless. RT kernel just adds some RT features to Linux, like guaranteed minimum execution slices and latencies. If your music doesn't skip while using the regular kernel, you can be pretty sure it won't skip either on the RT kernel. But if it doesn't skip in the first place...
In the case at hand ... mmmmmh, it would probably be useless. RT kernel just adds some RT features to Linux, like guaranteed minimum execution slices and latencies. If your music doesn't skip while using the regular kernel, you can be pretty sure it won't skip either on the RT kernel. But if it doesn't skip in the first place...
Well,
I don't know... this is audio, where improvements have no limits 🙂
Reports say that at 352/384K sample rate, it skips. Either the processor is running out of juice or some other issue...
I was thinking more in terms of assigning priorities to the different processes, some being RT priority (but have very little experience/knowledge in this respect)
I don't know... this is audio, where improvements have no limits 🙂
Reports say that at 352/384K sample rate, it skips. Either the processor is running out of juice or some other issue...
I was thinking more in terms of assigning priorities to the different processes, some being RT priority (but have very little experience/knowledge in this respect)
Seem to remember reading something to that respect. Could have been running the software SRC. I'll have to review the whole thread 🙂
- Home
- More Vendors...
- Twisted Pear
- Building an open embedded audio applicance.