ASIO Buffer size

[PaperBoat]

Please help me to understand Buffer and Sample size functions in ASIO devices.

-------------------------------------------------------------------------------------------First scenario-------------------------------------------------------------------------------------

My Foobar2K buffer is set to 50ms. ↓

My IFI USB audio device buffer size is set to 8 samples. ↓

In this scenario, Foobar is able to play music without any glitches.







-------------------------------------------------------------------------------------------Second scenario---------------------------------------------------------------------------------



My JRiver buffer is set to "Minimum Hardware Size". ↓

My IFI USB audio device buffer size is set to 8 samples. ↓

In this scenario, JRiver is failed to play music.










----------------------------------------------------------------------------------------------------Third scenario--------------------------------------------------------------------------



My JRiver buffer is set to "Minimum Hardware Size". ↓

My IFI USB audio device buffer size is set to 128 samples. ↓

In this scenario, JRiver is able to play music without any glitches.

Note: JRiver is unable to play music if IFI USB audio device buffer size is set to below 128 samples.

What is going on?

 

[Tom C]

I am hoping you get some good answers here.

 

[DVDdoug]

A buffer is like a storage tank or a long pipe. Buffers are required because of the multitasking operating system, and your OS is always multitasking even when you're running only one application. A buffer is also a delay.

Data is written to the output/playback buffer in a quick burst and then it flows-out to the DAC at a smooth-constant rate. If the buffer isn't re-filled in time you get buffer underflow and a glitch/gap in the audio. This happens if the buffer is too small or if something hogs the system for a few-milliseconds too long. Whatever is interrupting the audio doesn't have to be using a lot of total CPU time, it just has to hog it for a few extra milliseconds.

There is also an input/recording buffer that works the opposite way. The audio data flows-in smoothly and is written to the hard disk in a quick burst when the system gets around to it. If the buffer is not read in time, or if the buffer is too small, you get buffer overflow and a glitch.

With a bigger buffer you get more latency (delay) and less chance of a glitch.

Latency is mostly an issue while recording and monitoring yourself on headphones. It's hard to perform with a noticeable delay in your headphones. Or it can be a problem if you are watching a video and the audio & video latencies don't match and you get a "lip sync" problem. For everyday listening, a few (or several) milliseconds of latency is not a problem. I've got recordings with 60 years (or more) of "latency".

172 samples at 44,100 samples-per-second is 0.0039 seconds = 3.9 milliseconds.

 

[AnalogSteph]

Foobar2k internal "Buffer Length" has nothing to do with the ASIO driver buffer, which is the setting that JRiver seems to respect. You should be able to open the ASIO driver settings dialog from within Foobar, probably under "More options". Its settings are generally maintained per application.

BTW, ASIO is not generally worth messing with in a playback application these days. WASAPI exclusive mode tends to be just about equivalent.

 

[PaperBoat]

Foobar2k internal "Buffer Length" has nothing to do with the ASIO driver buffer, which is the setting that JRiver seems to respect.

Is there any benefit when using EQ DSP in players with lower software buffer? Like JRiver is offering lower buffer setting than Foobar2000... I'm asking because I think changes in EQ DSP will more effective with lower buffer samples... Isn't it?

 

[voodooless]

I'm asking because I think changes in EQ DSP will more effective with lower buffer samples... Isn't it?

What do you mean by “more effective”?

EQ works exactly the same, regardless of any buffer size. There may be some advantages with regards to CPU usage with various buffer settings, but the outcome, except for delay is the same.

 

[HarmonicTHD]

For just play back, latency is not relevant. Set the buffer as large as you can. (Latency becomes important when recording eg playing an instrument while listening to yourself playing while you either record the instrument or run it through effects eg DAW).

 

[PaperBoat]

What do you mean by “more effective”?

EQ works exactly the same, regardless of any buffer size. There may be some advantages with regards to CPU usage with various buffer settings, but the outcome, except for delay is the same.

Maybe my wrong choice of words. But I think when working with bigger samples DAC don't need to request information for a while... In the meantime changes that made by EQ DSP will have to waiting for next big samples transmission.

 

[PaperBoat]

For just play back, latency is not relevant. Set the buffer as large as you can. (Latency becomes important when recording eg playing an instrument while listening to yourself playing while you either record the instrument or run it through effects eg DAW).

The audiophile world is full of confusing things!

 

[voodooless]

Maybe my wrong choice of words. But I think when working with bigger samples DAC don't need to request information for a while... In the meantime changes that made by EQ DSP will have to waiting for next big samples transmission.

That’s not how it works. These buffer sizes are largely independent from how the DAC requests data over the USB bus.

 

[PaperBoat]

That’s not how it works. These buffer sizes are largely independent from how the DAC requests data over the USB bus.

Okay... But I would like to know what is "DAC Link"? Is it a Voodoo thing?

 

[voodooless]

It’s just something to control latency. Unless you do live audio mixing, I would not care about the setting. It has no influence on sound quality, other than that maybe some settings won’t give stable playback m.

 

[PaperBoat]

"Some people prefer high values and others prefer low."

That's drives me crazy...

 

[voodooless]

"Some people prefer high values and others prefer low."

That's drives me crazy...

That’s just nonsense, ignore it.

 

[PaperBoat]

That’s just nonsense, ignore it.

Thanks for confirming it.

 

[jhwalker]

"Some people prefer high values and others prefer low."

That's drives me crazy...

... because they don't know any better.

As other posters have noted, buffer size / latency is totally irrelevant to sound playback "quality", other than larger buffers perhaps making it less "glitchy" in busy system.

It may only become noticeable if you are doing real-time recording / monitoring.

 

[PaperBoat]

Don't worry I've many of them ↓

Our experience is that a small buffer sounds better.

 

[voodooless]

Don't worry I've many of them ↓

Our experience is that a small buffer sounds better.

Alpha Audio

 

[dfuller]

Buffer size only has to do with latency vs stability. Most people who need to worry about that are recording, with monitoring from the DAW so as low a RTL as possible is desired.

For just listening, it makes absolutely no difference.