24 bit vs 16 bit for Jajuk

[davidsrsb]

I have been ripping cds to 44100/16 bit flac files for my PC with Asus Xonar Essence and am thinking that I should convert these files to 44100/24bit flac, which should be basically lossless, so that the digital volume control has enough bits to work on without causing significant rounding errors.

The PC is running Ubuntu 10.10 and 100% volume causes distortion

Anyone with any thoughts about this?

[phofman]

The volume control is provided by digital circuits of the DAC chip in your sound card. It operates on the same data length regardless of the input audio file format - lower-precision formats have lower bits padded with zeros in the driver.

[Gopher]

You won't be 'converting' anything to 24 bit. The information just isn't encoded on the original CD to do that. You'll be guessing the intermediate values - that's all.

[davidsrsb]

Agreed that I cannot add resolution. I am thinking that the card is capable of 24 bit (20 bit in practice) resolution, so if I set the volume to about 0.7 to avoid distortions, I don't want the OS software multiplying a 16 bit value by some constant and truncating to 16 bit again.

A -60dB sinewave is already horribly quantised in 16 bit format. If the card actually is working as 24 bit, then no problem as this waveform will be fairly accurately scaled.

Isn't the volume control implemented in the main cpu, not in the DAC these days?

[phofman]

Originally Posted by davidsrsb

Agreed that I cannot add resolution. I am thinking that the card is capable of 24 bit (20 bit in practice) resolution, so if I set the volume to about 0.7 to avoid distortions, I don't want the OS software multiplying a 16 bit value by some constant and truncating to 16 bit again.

Even if the volume control was done in software (e.g. fade-in/out in modern players), the software uses its internal sample format (mostly float or int32 or int16 for older programs). The input format is converted to the internal format first and futher processing is independent of the original sample length.

Originally Posted by davidsrsb

Isn't the volume control implemented in the main cpu, not in the DAC these days?

It all depends on your playback chain. It can either use software volume regulation (softvol plugin of alsa, software volume in mplayer when configured, software volume in pulseaudio when configured), or the volume control provided by the driver and operated via corresponding alsa-lib API (mplayer when configured, pulseaudio when configured, various mixer gui applets). The driver transmits the commands further to the sound card (DAC chip or analog volume regulation chip on old cards or Revo5.1 ).

[Max 002]

First of all, isn't this card running in 32 bit? Anyways i don't think your bit rate is the problem...

Most low-end equipment will distort even before the 0dbfs limit. Thats why mastering engineers often export their mixes a -0.5dbfs. So pushing your card's output over that threshold is probably causing the problem.

Second of all, i don't understand why you need to set the output so loud! I never tryed to put my volume that loud cause is was afraid to kill my headphones. (Im using the beyerdynamics DT 770 pro 80Ohms. Good headphone listening level is around -20dbfs. And if you need higher volume in the speakers, put the volume up on the amp.

I hope this helps!

-Max

[davidsrsb]

There was a time when it was the high end CD players with custom dsp filters that overloaded at 0dbfs.
Unfortunately many of my recent CD clip to full scale