16bit sources on 32bit USB receiver

General Discussion about MPD – anything that doesn't fit in the other MPD forums.
Post Reply
davor281
Posts: 1
Joined: May 5th, 2013, 9:40 pm

16bit sources on 32bit USB receiver

Post by davor281 »

I found this on well tempered computer -

Will playing 16 bits sources on a 24 DAC affect sound quality?
Very unlikely, all what is done is padding 8 zero’s too the 16 bits to make it a 24 bit word.
In fact there might even be a benefit, if you apply digital volume control, you can chop off these eight zero’s (48 dB reduction) without losing resolution.

I own a Audio-gd NFB-17.32 DAC which has a 32bit receiver and Wolfson 8741 can accept 32bit data. On the Audio-gd site it is recommended to set the bit transfer to 32bit for better quality.

I am using mpd and I would like to have some opinions about setting the bit transfer to 32 in mpd.conf file? (adding format "*:32:*" line)

Are this just added zeros or does this acctually helps in better performance?

Thank you,
Davor.
Ran
Posts: 151
Joined: February 25th, 2013, 3:47 am

Re: 16bit sources on 32bit USB receiver

Post by Ran »

You should not try to up sample your source. Let your DAC do that for you. Also, padding a 16 bit source file will add nothing but digital noise. If you have a 16 bit file (redbook), let your DAC play it as is. Up sampling is tricky and many DACs perform this in various ways. Some Hi-End DAC do not do that at all. The best setting in MPD is when "bit perfect" is enabled. You should try to by pass the operating system's mixer and use the hardware output when possible.
Qball
Forum team
Posts: 54
Joined: January 15th, 2013, 4:03 pm

Re: 16bit sources on 32bit USB receiver

Post by Qball »

I agree partially with ran, if you have a good! DAC do not change bit depth.
But like always, there is a but:

1. Some cheaper DACs perform significantly better at 24bit then 16bit. They might not automatically upsample.

2. When doing Software Volume (so modifying the sound to reduce volume), use 24/32 bit.

Also padding with 0's should not add digital noise.
Post Reply