Author Topic: Clarification for Output Resolution and Bitrate  (Read 2580 times)

Zzyzx

  • Posts: 55

After reading through a whole lot of posts, I just wanted ask for clarification on the Output Resolution and Output Sample Rate settings to make sure I've got this right (the settings on the Output dialog of Options and Stuff).

As I understand it...

The Output Resolution (bit depth) sets specifically what XMPlay will output.  Regardless of what the original file format is.  If the Output Resolution is less than the original file format, there will be a down conversion done.  If it's higher, what exactly does it do, just pad the extra bits?

However, if any DSP is used (including the standard volume or graphic equalizer) then XMPlay will convert the original file format to 32 bit floating point to apply the DSP algorithms to it.  Then there is always a conversion to the Output Resolution.  So, it would be best (assuming the output driver and audio device supports it) to set the Output Resolution to 32 bits to get the best possible result from the 32 bit floating point internal format conversion.

The Output Sample Rate (bit rate) is only used if the Apply Sample Rate to All File Formats is checked.  Otherwise the Output Sample Rate is ignored and the bit rate of the original file is used.

Is my understanding of the settings and DSP bit rate conversions correct?

So, I am using the built-in volume and graphic equalizer DSPs.  I am outputting via ASIO to my ASUS Xonar Essence One, which supports 16, 24 and 32 bits depth, and 44.1, 48, 88.2, 96, 176.4 and 192 KHz bit rates.  The ASIO driver is "bit perfect" and handles all those formats.  It looks to me like I should uncheck Apply Sample Rate to All File Formats so the bit rate is always the same as the original file format, and set Output Resolution to 32 Bit to get the best conversion rate from the internal 32 bit floating point DSP format used.

Doing that, and looking at the Info/General window while playing, it looks like XMPlay is always showing the Output to be:  "<original file's bit rate> - stereo - 32 bit".  Which I think is what I want for the best possible output.

Do I have this right, or am I mistaken?

Thanks for any clarifications.  Sorry for the wordiness of this posting -- I wanted to make sure I was clear.

amit

  • Posts: 723
Re: Clarification for Output Resolution and Bitrate
« Reply #1 on: 17 Nov '12 - 09:50 »
As I understand it , all files are initially transformed into 32 bit and all calculations (pre-amp,dsp,volume) are done with the added accuracy. Then at the last stage the audio stream is truncated to the desired bit depth (after dithering or without it).

Sample rate is a different aspect of the audio stream that is not related to bit depth. I am guessing a change of it is done , if requested , just prior to the last bit depth change.

Dotpitch

  • Posts: 2871
Re: Clarification for Output Resolution and Bitrate
« Reply #2 on: 17 Nov '12 - 16:25 »
Do I have this right, or am I mistaken?
Yes, you're right. And how does it sound?

The Output Sample Rate (bit rate) is only used if the Apply Sample Rate to All File Formats is checked.
Note that sample rate is not the same as bit rate. The sample rate refers to the number of samples per second (for example 44.1 kHz), while the bit rate refers to the number of bits used to encode one second of audio (for example 128 kbit/s).

... all files are initially transformed into 32 bit and all calculations (pre-amp,dsp,volume) are done with the added accuracy. Then at the last stage the audio stream is truncated to the desired bit depth (after dithering or without it).
Actually, decoding the audio results in floating point values, rather than 32-bit integer values. Conversion to integers (for the soundcard output) is indeed the last step.

Zzyzx

  • Posts: 55
Re: Clarification for Output Resolution and Bitrate
« Reply #3 on: 17 Nov '12 - 17:16 »
Yes, you're right. And how does it sound?

I think it sounds pretty amazing, especially for the high resolution music I have (88.2/96/192 KHz 24 bit).  The ASUS Xonar Essence One is a nice DAC unit, especially when fed into a higher end audio system.  I'm not sure I can actually notice all the nuances in A/B tests, but I just wanted to set it up with the theoretically best options.   ;D

Note that sample rate is not the same as bit rate. The sample rate refers to the number of samples per second (for example 44.1 kHz), while the bit rate refers to the number of bits used to encode one second of audio (for example 128 kbit/s).

Yes, you're correct, thanks, my terminology is off there.

Thanks for the confirmations!

winner

  • Posts: 260
Re: Clarification for Output Resolution and Bitrate
« Reply #4 on: 20 Nov '12 - 00:34 »
Yeah, the narrative here seems correct to my thinking as well, after having grappled over time with understanding the various signal changes that occur in transit through XMPlay and various plugins.

I do often use Winamp plugins that require a 16-bit output resolution. Of course I'd rather have 24- or 32-bit final output, but some of those plugins are just too cool or useful to dismiss for that reason.

In practical use though, I don't think most people in common listening circumstances can hear much of a difference between quality-produced sounds, whether at 16-, 24-, or 32-bit depth.

Zzyzx

  • Posts: 55
Re: Clarification for Output Resolution and Bitrate
« Reply #5 on: 20 Nov '12 - 19:32 »
In practical use though, I don't think most people in common listening circumstances can hear much of a difference between quality-produced sounds, whether at 16-, 24-, or 32-bit depth.

When it comes to some differences, I'd tend to agree.  For example, the difference between 96 KHz and 192 KHz audio at the same bit depth doesn't seem to be noticeable to me.  I'm sure with my 52 year old hearing, I can't hear much above 10K any more anyway, but even if the higher harmonics are more sensed than heard, I haven't been able to reliably discern between them in double-blind A/B tests.

Also, I'm suspect of some of the 192 KHz audio out there (for example, the 192 KHz albums sold by HD Tracks).  I've read that some of the 192 KHz audio is actually up-converted from 96 KHz 24 bit masters that were made from the original analog tapes, so I'm not sure what I'm actually getting sometimes.  I wish they would clearly state in which format the original digital master was done.

Now given real good quality audio source material, for example something like Steely Dan's Two Against Nature, I can very reliably tell the difference between the 44.1 KHz 16 bit version, and the 96 KHz 24 bit version in double-blind A/B tests.  And that's not with ultra-high-end equipment (XMPlay to ASIO to ASUS Xonar Essence One DAC to Klipsch ProMedia 2.1 system [old style ProMedia amp, not the newer crappy one]).  There is just so much more "life" in it, it seems pretty obvious to me.  I can also really tell the difference between Steve Hackett's Genesis Revisited II in 88.2 KHz 24 bit vs. the 44.1 KHz 16 bit CD version.

I really think the bit depth makes a difference in the dynamic range, so the difference between 16 bit and 24 bit to me is noticeable.  That's a 6 dB range difference, assuming it's actually used in the source material (i.e., the original source was 24 bit in the first place).  Unfortunately, a lot of contemporary source material is so compressed and limited that they squeeze the life out of it and the extra bit depth does nothing for it.

I know some folks would consider this a bad thing, but I've had a lot of success using a declipping program (I use SeeDeClip Pro -- http://www.cutestudio.net/data/products/audio/seedeclip/index.php) to restore the limited peaks.  When I do that I increase the bit depth (16 to 24, or 24 to 32) so it effectively increases the dynamic range for the restored peaks.  It's very noticeable, and I think for the good, although like I said some folks would consider it altering the material, which it is of course although I would argue it's more of a restoration.
« Last Edit: 20 Nov '12 - 19:36 by Zzyzx »

saga

  • Posts: 2181
Re: Clarification for Output Resolution and Bitrate
« Reply #6 on: 20 Nov '12 - 20:32 »
Everything above 16-Bit is mostly irrelevant for normal music listening, but more relevant to audio editing (including mixing / mastering). I doubt that you can hear a difference between a downmix of the same piece in 16-Bit and 24-Bit. Plus, I don't think that you'll find any DACs and ADCs that can actually input/output more than 24-Bit. 32-Bit floating point audio is a bit of a different story again, since it has the same dynamic range as 24-Bit fixed point audio.