Music Creation On Pandora

[Snu]

The processor and DAC support audio output at up to 24 bits/ 48KHz although most people will likely stick with 44.1KHz, 16 bit to match common emulator audio and MP3 playback, but the option is available. If you need 24 bit/192KHz, USB is the only way to go

Audio recording is not supported by Pandora natively. It wasn't a justifiable cost. Personally, I would rather plug in an M-Audio recorder or similar if I wanted to record quality stuff.

doesnt it have a mic input? or is that 22khz 8bit mono or something?
figured i would get a usb soundcard for input and midi ports, i'll wait until pandora is out and we are sure of which devices work before dropping some money on that tho...
the 24bit playback is awesome news (and should be added to the wiki immediately!), anything over 48khz isnt an audible difference (for me), but its pretty easy to hear the difference in the noise floor between 16bit and 24bit.

 

[MWeston]

doesnt it have a mic input? or is that 22khz 8bit mono or something?

Yes, there are two mic inputs (one internal, one external) but they are voice grade. I think 16KHz, 16 bit or something like that. They will work well for in game chat, skype and taking notes, but I sure wouldn't record your band jamming with it.

 

[M-.-n]

But you can have usb one right ? please don't tell me audio & midi usb will not be supported ... pretty please?

Yes, there are two mic inputs (one internal, one external) but they are voice grade. I think 16KHz, 16 bit or something like that. They will work well for in game chat, skype and taking notes, but I sure wouldn't record your band jamming with it.

 

[crazedover]

But you can have usb one right ? please don't tell me audio & midi usb will not be supported ... pretty please?

I think as long as you have the linux drivers for those devices you'll be fine.

 

[Snu]

But you can have usb one right ? please don't tell me audio & midi usb will not be supported ... pretty please?

have you even read this topic?

 

[M-.-n]

Well I thought I did, but obviously I blanked out some parts... sorry for the garbage.

Exellent news, this could take littlegptracker to some new grounds.

But you can have usb one right ? please don't tell me audio & midi usb will not be supported ... pretty please?

have you even read this topic?

 

[A_SN]

i would LOVE if it was capable of 24bit input and output, even at only 44 or 48khz... but thats pretty doubtful.

I can easily understand why you'd want a 24 bit input, although it wouldn't magically make things be any less noisy, but there's no possible reason you could appreciate 24 bit output. The dynamic range of 16 bit audio is 96 dB. Not only you can't possibly theoretically appreciate anything better than that, but you do not have a system that has such a low noise level.

As for anything above 44 khz, you're not a dog or a bat, so you can't appreciate anything better either. If you don't believe me just read about the Shannon-Nyquist theorem and human hearing range.

 

[Snu]

I can easily understand why you'd want a 24 bit input, although it wouldn't magically make things be any less noisy, but there's no possible reason you could appreciate 24 bit output. The dynamic range of 16 bit audio is 96 dB. Not only you can't possibly theoretically appreciate anything better than that, but you do not have a system that has such a low noise level.

As for anything above 44 khz, you're not a dog or a bat, so you can't appreciate anything better either. If you don't believe me just read about the Shannon-Nyquist theorem and human hearing range.

im well aware that im unable to hear frequencies over 18khz or thereabouts (i havnt done an actual test, so im not sure the exact number, maybe its 16khz, maybe its 20... well, probably not).
however, there are numerous arguments about aliasing when it comes to higher than 48khz recordings (tho, i wont be arguing those, as i mentioned i cannot hear any differences). also, personally i dislike 44khz simply because its an 'odd' number which came about due to hardware limitations rather than logical reasons.

and as i mentioned, in my experience there is certainly an audible difference between 16bit sampling and 24 bit sampling. i havnt done any research on that, so i cant argue specifics, i can only say what the results of my tests have been. perhaps it has to do with aliasing as well, i dont know.

 

[M-.-n]

and as i mentioned, in my experience there is certainly an audible difference between 16bit sampling and 24 bit sampling. i havnt done any research on that, so i cant argue specifics, i can only say what the results of my tests have been. perhaps it has to do with aliasing as well, i dont know.

As far as I could grasp it, the most important with respect to "sampling frequency" is the internal computation. A filter treatment will allow a lot more subtleties if run at higher frequencies... of course, it means better resulution/accuracy but true 'analogish' modelisation needs very high frequency rate to be accurate. Most audio engine "relate" their input/output frequencies to the internal engine resolution but I'm pretty sure running in 44.1 for output would be fine if all in internal path where at 96...

Just a thought,
Marc.

 

[scarath]

But you can have usb one right ? please don't tell me audio & midi usb will not be supported ... pretty please?

I think as long as you have the linux drivers for those devices you'll be fine.


On my debian box I am using this USB sound card, worked instantly, etc. So assuming it works just as well there shouldnt be a problem with getting good sound out of the Pandora.

 

[A_SN]

however, there are numerous arguments about aliasing when it comes to higher than 48khz recordings (tho, i wont be arguing those, as i mentioned i cannot hear any differences). also, personally i dislike 44khz simply because its an 'odd' number which came about due to hardware limitations rather than logical reasons.

Usually these arguments are brought up by "audiophiles", the type who buy gold cables and wooden knobs for a "warmer sound" or even put their CDs in the freezer so that the sound sounds "softer", and are debunken by engineering types.

As for the sample bit rate, it's really all a matter of noise/dynamic range.

As far as I could grasp it, the most important with respect to "sampling frequency" is the internal computation. A filter treatment will allow a lot more subtleties if run at higher frequencies... of course, it means better resulution/accuracy but true 'analogish' modelisation needs very high frequency rate to be accurate. Most audio engine "relate" their input/output frequencies to the internal engine resolution but I'm pretty sure running in 44.1 for output would be fine if all in internal path where at 96...

Well the thing is, not all programs/hardwares use ideal interpolation/filtering techniques, and for the sake of computational efficiency they tend to use relatively short windowed sinc functions, and the shorter they are the larger the frequency range they take to "fade out" is. (EDIT : Hence the need for some "head room")

But you can obtain theoretically perfect results (with respect to your "analoguish modelisation" remark) with as low as 44.1 kHz, and matter of fact if you can't hear a thing above 16 kHz (which is the case for lots of adults), then you can obtain perfect results with just a 32 kHz sampling rate. So really, there's no theoretical need for any sampling rate higher than twice the highest frequency one can hear, only practical reasons.

 

[M-.-n]

But you can obtain theoretically perfect results (with respect to your "analoguish modelisation" remark) with as low as 44.1 kHz, and matter of fact if you can't hear a thing above 16 kHz (which is the case for lots of adults), then you can obtain perfect results with just a 32 kHz sampling rate. So really, there's no theoretical need for any sampling rate higher than twice the highest frequency one can hear, only practical reasons.

You're talking about the direct 'result' ... i.e the sound stream output and for that I pretty much agree (although 32k is pushing it) but you can have higher order effects that do have a result in the audible range (i.e. 44.1 in our case) but don't 'propagate' properly unless they are computed at very high frequency.

 

[A_SN]

You're talking about the direct 'result' ... i.e the sound stream output and for that I pretty much agree (although 32k is pushing it) but you can have higher order effects that do have a result in the audible range (i.e. 44.1 in our case) but don't 'propagate' properly unless they are computed at very high frequency.

Like???

 

[M-.-n]

Like???

It's out of my reach really but I know having a long discussion with a guy way into filter modelling and that's what he told me. It's related to non-linearities in the process but I won't be able to explain that to you myself.