Re: arecord command line options

On Tue, 29 Sep 2020, Ralf Mardorf wrote:

> On Mon, 28 Sep 2020 22:02:10 -0400, Alan Corey wrote:
>> On 9/28/20, Zsolt Ero <[email protected]> wrote:
>>> I have a few questions related to arecord, which I couldn't find in
>>> the man pages nor anywhere on the internet.
>>>
>>> My use case is very simple, I'd like to record stereo 24/192 audio
>>> into a WAV file from an E-MU 0202 USB interface using an old laptop
>>> (Core 2 Duo).
>>>
>>> For this I thought of using the device which starts with "hw:", as it
>>> seems to me that it provides the least processings over it. The
>>> recording works without problems, however I'm confused about the
>>> options.
>>>
>>> What is not clear to me:
>>> --mmap - should I use it or not? I'd be writing from USB to hard
>>> drive a wav file.
>>> --period-time, buffer-time, period-size, buffer-size - I'm totally
>>> confused about these. I'd like the highest possible recording
>>> quality, latency doesn't matter to me.
>>> --avail-min - what is a wakeup?
>>> --disable-resample/channels/format/softvol - do I need this if I
>>> selected the hw: device? I'd like to record without any kind of
>>> processing.
>>> --test-position/coef/nowait - do I need this for my use case?
>>>
>> I would say you're in danger of overthinking it, try the defaults
>> first.
>>
>> Except when you say 24/192 I'm thinking 192 kbits/sec?  My usb sound
>> cards can only do 48 max.  Yours may do better.  I was just looking
>> into it for SDR purposes, the tuning range depends on the max sample
>> rate.  SPDIF has always seemed like the way to go, or maybe firewire.
>> But you need to study the E-MU 0202 specs to be sure it can do 192.
>
> It's for 24 bit, 192000 Hz sample rate. FWIW professional
> studio audio recordings are usually done at 48000 Hz.
> A sample rate > 48000 Hz doesn't gain better audio quality. To increase
> audio quality you need to replace the prosumer by a professional audio
> interface. Even if the E-MU 0202 USB should use good converters, it
> much likely suffers from a not that good analog component. Loosely
> speaking, recording frequencies that are inaudible for humans and/or not
> playable by analog audio equipment, doesn't increase the audio quality.
> It's common to cut frequencies that are inaudible for good reasons.
> When taking photos or when filming you don't increase the image quality
> by including ultraviolet light, actually it's better to get rid of
> ultraviolet light.
> Again loosely speaking, the higher the bit rate, the better, but very
> high bit rates are just an advantage for further processing, not for
> just playing and listening.
> In short, with your device you much likely get the best result with 24
> bit, 48000 Hz. When using a sample rate > 48000 Hz you unlikely will
> benefit from a possible advantage, you more likely will suffer from
> a possible negative side effect
>
>

The possible negatives is that the high frequencies dumped into your audio
equipment, including speakers, could trigger non-linearities at thos high
frequencies which then mix down to the audio band, expecially if those high
frequencies have an appreciable amplitude. And even 24 bit never is. The noise
floor of the electronics etc is considerably higher than that. So you might
get 18 bit effective out of the 24 bit if you are lucky (ie, a slight increase
in dynamic range). 
Ie, the difference between 24/192 and 16/44.1 at best is miniscule, and at
worst, worse in the first case.
But since you have the E-MU you might as well use it.