|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
101.113.46.109
Hi,
When you're using something like Audirvana and it says output is 32Bit and say 44.1kHz, feeding a 16Bit DAC chip, does that mean I could use up to 16dB attenuation without losing any resolution?
Thanks for any comment.
Shane
Follow Ups:
If the DAC is 16 bits, it is 16 bits.
Makes me wonder what happens if Aurdirvana outputs 32 bits.
It probably does all of its processing in 32 bits to improve precision but in the end the results must be truncated to 16 bit otherwise this DAC won’t play.
On a 16 bit DAC you will loss 1 bit for each 6 dB reduction.
If you play 16 bit audio on a 24 bit DAC you can (theoretically) lower the volume by 8*6= 48 dB before you lose resolution.
The Well Tempered Computer
So lets say its Audirvana and its playing back a 16bit file to a 16 bit DAC.
Where it says 32Bit/44.1kHz output, is it creating an additional 16 bits of data ?
The DAC chip can only process 16Bits and truncate any additional, so the question is, any reason why those extra 16bits cant be used for digital attenuation?
I guess it all comes back to whether 32/44.1kHz actually means its outputting 32bit word length from a 16Bit source.
Any ideas?.
Assuming the application works correctly and is set to output 32 bits, then when playing a 16 bit file at 100% volume it would output the high order 16 bits and fill all the remainder at zero. It would be a mistake to configure the ap to output at 32 bits with a 16 bit DAC. So I would assume that the DAC accepts 32 bit input. If the DAC works correctly then it will produce the same results on a 32 bit input with 16 low order zeros as it would have with a 16 bit input.If the application applies digital volume control (or upsampling) the resulting processing will add additional low order bits depending on what was done. If the output was set to 32 bits then there would be no loss of resolution on volume control until 16 bits of digital volume control were applied, which would amount to a volume reduction of -96 dB. So in this case no resolution would be lost in sending data to the DAC.
Of course the DAC almost certainly doesn't have 32 bit resolution, but that's a different question. The details of what you hear will depend on the DAC. It could be that what you hear will be better with a lower digital signal into the DAC instead of sending a full volume digital signal into the DAC and then reducing it downstream by an analog volume control. This is the situation with my Mytek DAC. Here I use a built in digital volume control in the SABRE chip that works at 32 bits. There is no loss of digital resolution at the settings I typically use, e.g. -16 dB, starting with 24 bit sources. There is more gain out of the DAC and so any noise will be amplified. However, I can't hear any noise out the speakers, even if I put my ears right at them. So there is no real loss of signal to noise resolution going this way. Actually, the music sounds better, because the op-amps in my DAC have less distortion when they are working at a lower output level.
If you are using a 16 bit DAC then you should not use a digital volume control, because you need to use all the limited resolution how have. Given that there are DAC chips that provide higher resolution why would any one want to use an antiquated 16 bit DAC chip? If one subscribes to the NOS religion there are DAC chips that work at higher bit depths.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Edits: 07/22/14
Thanks for the reply Tony.
The DAC chip wont accept 32bits, this much I know.
Audirvana says 'WAVE 16/44.1kHz : DAC32/176.4kHz'. I cant seem to change that.. I should say that the software actually see's a 32Bit 384kHz capable device in Lucians WaveIO.
This feeds Ians FIFO then ISO and Dual Reclock boards into his I2S to PCM board which allows simultaneous mode operation of tda1541A in NOS at 176.4kHz, upsampling using the Audirvana algorithm.
So, I'm assuming that the file is output from the software at 16bit with another 16 positions held at zero for 32, which is passed through the chain to the DAC chip with discards the 16 zero's and playsback the 16bit file.
If that is correct, is it still the case that you would suggest not to use digital attenuation?. Maybe, in other words.. would the attenuation affect the 16 zero's first - or is that irrelevant?.
As for subjective preferences, theres already so much mentioned about this elsewhere that its pointless to question or even comment about it if one is to remain without agenda and for the most part ego.
Sincerely,
Shane
I would sincerely hope that dither and/or noise shaping rather than truncation would be used to to go from higher bitrates to 16.
Truncation is primitive, inelegant and a serious compromise of sound quality as it creates quantization distortion.
It depends on the media player and let’s hope most of them do it in a graceful manner.
However, if you have a media player processing in 32 bit (or 64 as in case of JRiver) in the end only 16 bits can be send to a 16 bit DAC.
Converting to 16 bit integer instead of truncating is a more accurate description.
The Well Tempered Computer
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: