Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
I'm doing some experiments with buffer settings in JRiver.
My setup is a MacBook Pro with an iFi Nano iDSD DAC (custom modified by an acquaintance). I've tried both USB 3.0 and Thunderbolt external drives to hold the music.
My software is JRiver Media Center.
So when I first starting using JRiver it seemed like lower buffer sizes were better. I'm referring here to four settings:
- software buffer
- hardware buffer
By the way that's a hell of a lot of options and settings if I was to actually try all combinations, but I'm just trying things and listening a lot.
So basically lower buffer settings seemed to offer more speed and extension, while higher buffer sizes sounded a bit constrained dynamically and congested.
But then everything changed. I discovered the setting in JRiver called "exclusve access to sound device." When I checked that, first of all the sound got much better. Secondly, now it seemed like larger buffer sizes were sweeter and smoother with a lower noise floor. There was no sense of constraint or congestion any more with the larger buffer sizes.
Another option to play with is dithering. I'm a bit confused what JRiver actually does when you turn on dithering. I thought that dithering was handled by the recording engineer in producing the original file, or maybe the DAC does something like dithering in playback, but I thought the point of the media player was to send the data to the DAC in bit perfect form.
(oh i should say at this point that most of my files are 44.1/16. My favorite recordings are usually available only on red book CDs)
Nonetheless, changing the dithering setting has an effect. It seems to vary with the recording. Sometimes I feel like turning on "JRiver Bit Perfect Dithering" improves leading edges. The drums really pop and slam. Amazing how I felt like the drummer was right in the room with me. But on other music, it seems that turning dithering off gives a more neutral and truthful ("objective") perspective which can often really improve the musical expression -- the musician's point gets through better.
You are correct and I have been saying for a long while that there are so many settings (not only in the player but elsewhere too) that it is near impossible to find an optimal setting for all kinds of music and hradware. Those who buy several 'optimisation' programs and use them on top of each other actaully end up not knowing what has been done to their opertaing system and audio drivers.
If you are using xmos uab software in a pc system, you are running with a usb processor buffer size which has been set as 'safe' by xmos. However, if you are lucky to have a device with Theyscon drivers, then you will find that settings in the usb CPL affects SQ signficantly as well.
And so I maintain that there is no such thing as audio optimisation that applies scross software and hardware platforms. Fiddling to ones satisfaction by ear is really the terminology to be used!
Fiddling to ones satisfaction by ear is really the terminology to be used!
I agree, at least for those who are brave enough to fiddle with things. You'd have to be brave to trust your ears, because they might just like something that does not meet audiophile standards.
that those who cannot tell differences are possibly listening to rock and other music with a relativelysmaller dynamic and tonal range.
Most older adults have some amount of hearing loss
This doesn't negate perception of good audio by way of imaging, individual instrument placement, sound stage size etc which are very evident when systems are 'improved'.
I am in full agreement with all that you've said! Very nicely worded.
That is consistent with what I've learned and experienced with my mentors. :)
Maybe, maybe not... Often these things friend on subtle cues which may go unheard.
PCM audio is integer (16 or 24)
Likewise our DACs only accept integers (16/24/32)
Any type of DSP, be it down-sampling from 24 to 16 or volume control, is a calculation.
There will always be a remainder (1/3=0.3333) and of course this won't fit into a integer.
Unfortunately, this remainder (quantization error) is correlated to the sound.
Say anything below 0.5 set the LSB to 0 and anything above to 1.
To get rid of this; dither is applied before the truncation.
It is a good practice to apply dither anytime DSP is used.
If you do "nothing" you are using Windows default (DS)
The audio is send to the mixer, converted to float, dithered and converted back to integer.
If you are using WASAPI, the same will happen.
If you are using WASAPI and "exclusve access" the mixer is bypassed. Hence you get rid of the dither as applied by Win.
If you use WASAPI in exclusive mode and all DSP in JRiver is disabled (volume control, DSP studio), you are sending bit perfect data to the DAC.
Bit Perfect Dither is nonsense by design, the moment one apply dither, one is no longer bit perfect.
My guess is , given the strong anti-audiophile sentiment at JRiver, this is their way out for applying dither with 24 bit material.
They probably think it nonsense to do so at -144 dBfs but public demand ask for it.
How do you set WASAPI to exclusive access using Jriver media center?
Tools > Options > Audio
Select the zone you want to configure
Select driver/WASAPI from the dropdown menu
In Device settings: check "enable device for exclusive access"
The Well Tempered Computer
It's known for a long time - at least to me - that buffers and buffer sizes can have certain impact.
And yes. The impact of large or small usually depends on each and every specific situation (SW and HW).
Sometimes this helps, sometimes that helps.
Usually dither got to be applied after the final mastering or DSP stage. Which usually would be digital volume control during playback.
And yes. This or that dither might interfere with the dither on the recording.
I'd fully agree with your conclusion.
I turned my dithered volume control off after some testing.
Post a Followup:
Post a Message!
This post is made possible by the generous support of people like you and our sponsors: