- Joined
- Dec 8, 2020
- Messages
- 3,051 (2.05/day)
- Location
- United Kingdom
Here is an interesting question I find difficult to answer myself. Going along the lines of a 32 bit float amp (PCM, software processing) and a CPU as the processor.
A 32 bit float amp can amplify up to 1,528 dB, which is insane, it should also be lossless amplification, far beyond any hardware amp.
How much THD, THD+N will the CPU add to the PCM audio (digital) during processing?
====
Currently you can send a sound system 144 dB (24 bit), but you will still need to add power (analogue style) to get something like 100 dB SPL output.
A 1:1 PowerDAC can accept 32 bit float input, then attenuate to meet its maximum power output, lets says 120 dB SPL, no amp.
The DMAS main unit is still the amp, so in this case, we input a 80 dB signal, then amplify to 100 dB (32 bit float).
----
Volume management can be a set decibel rather than random indicators, for example setting the volume to 100 dB (constant).
This way both 16 bit and 24 bit can be output at the same decibels, and a constant volume with all inputs.
It might also be worth adding set decibel per input, in case the user requires varying input levels.
====
In addition, all audio processing should be done in 32 bit float, and then attenuated to the set bit depth. This prevents digital clipping, and noise.
For example 144 dB > Sound effects (EQ for example) adds 6 dB, with 24 bit processing it will clip, and audio is lost.
If the processing was 32 bit float, the extra 6 dB is added without clipping or loss, then attenuated.
Low Level Signals: 32-bit Float versus 24-bit - Sound Devices
====
Also don't do a Microsoft and put volume management just before EFX, and allow SFX-MFX to clip (if 24 bit processing or less).
Edit, the level meter might be after MFX, but still before volume.
The one after SFX might per app, volume mixer.
Improvement: Master Volume/Per App > SFX > MFX > Level Meter > EFX.
App @ 144 dB > Master 80% (or app specific at 80%, same application point) [144 dB - 20% = 115.2 dB] > SFX|MFX [+ 10 dB = 125.2 dB] > EFX.
NOT: App @ 144 dB > SFX|MFX + 10 dB, clipped 144 dB > Master at 80% [144 dB - 20% = 115.2 dB] > EFX.
Ending the Windows Audio Quality Debate - (ASR) Forum
----
Also note Windows does technically support all digital formats with SPDIF, although you manually have to add them.
With Windows 11, ticking the additional formats will result in, DTS:X HT and Dolby Atmos in formats.
With Windows 10, ticking the additional formats will result in, Dolby Atmos in formats.
Potplayer, using WASAPI Exclusive - Event driven.
====
Reiteration in relation to the last post on page 1, and also SPDIF:
84 GBd (168 Gbit/s) PAM‐4 3.7 Vpp Power DAC in InP DHBT (2015)
An Ultrahigh-Speed Low-Power DAC Using InP HBTs (2010)
Note that TOSLink optical modules are rated in NRZ (PAM 2).
----
A 32 bit float amp can amplify up to 1,528 dB, which is insane, it should also be lossless amplification, far beyond any hardware amp.
How much THD, THD+N will the CPU add to the PCM audio (digital) during processing?
====
Currently you can send a sound system 144 dB (24 bit), but you will still need to add power (analogue style) to get something like 100 dB SPL output.
A 1:1 PowerDAC can accept 32 bit float input, then attenuate to meet its maximum power output, lets says 120 dB SPL, no amp.
The DMAS main unit is still the amp, so in this case, we input a 80 dB signal, then amplify to 100 dB (32 bit float).
----
Volume management can be a set decibel rather than random indicators, for example setting the volume to 100 dB (constant).
This way both 16 bit and 24 bit can be output at the same decibels, and a constant volume with all inputs.
It might also be worth adding set decibel per input, in case the user requires varying input levels.
====
In addition, all audio processing should be done in 32 bit float, and then attenuated to the set bit depth. This prevents digital clipping, and noise.
For example 144 dB > Sound effects (EQ for example) adds 6 dB, with 24 bit processing it will clip, and audio is lost.
If the processing was 32 bit float, the extra 6 dB is added without clipping or loss, then attenuated.
Low Level Signals: 32-bit Float versus 24-bit - Sound Devices
====
Also don't do a Microsoft and put volume management just before EFX, and allow SFX-MFX to clip (if 24 bit processing or less).
Edit, the level meter might be after MFX, but still before volume.
The one after SFX might per app, volume mixer.
Improvement: Master Volume/Per App > SFX > MFX > Level Meter > EFX.
App @ 144 dB > Master 80% (or app specific at 80%, same application point) [144 dB - 20% = 115.2 dB] > SFX|MFX [+ 10 dB = 125.2 dB] > EFX.
NOT: App @ 144 dB > SFX|MFX + 10 dB, clipped 144 dB > Master at 80% [144 dB - 20% = 115.2 dB] > EFX.
Ending the Windows Audio Quality Debate - (ASR) Forum
----
Also note Windows does technically support all digital formats with SPDIF, although you manually have to add them.
With Windows 11, ticking the additional formats will result in, DTS:X HT and Dolby Atmos in formats.
With Windows 10, ticking the additional formats will result in, Dolby Atmos in formats.
Potplayer, using WASAPI Exclusive - Event driven.
====
Reiteration in relation to the last post on page 1, and also SPDIF:
84 GBd (168 Gbit/s) PAM‐4 3.7 Vpp Power DAC in InP DHBT (2015)
An Ultrahigh-Speed Low-Power DAC Using InP HBTs (2010)
Note that TOSLink optical modules are rated in NRZ (PAM 2).
----
Attachments
-
32 bit float.png -
Low CPU Usage.png
Last edited:
