AdamGrossmanLG
Max Output Level: -62 dBFS
- Total Posts : 1436
- Joined: 2014/07/13 03:40:34
- Status: offline
Sampling Rate / Latency Correlation
Hello, I have been reading that the HIGHER than sampling rate, the LOWER the latency? That sounds wrong! Wouldn't a higher sampling rate cause the hard drive and CPU to work even harder thus making latency even longer? I am so confused as to what Sampling Rate i should be tracking at. Right now I am at 24bit / 44.1 (i see a lot of people do 48 Khz - not sure why as CD is 44.1 anyway). Any help here would be appreciated.
|
tenfoot
Max Output Level: -53.5 dBFS
- Total Posts : 2186
- Joined: 2015/01/22 18:12:07
- Location: Qld, Australia
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 01:53:29
(permalink)
As an example, with your asio buffers @ 512 samples doubling the sampling rate from 48khz to 96khz will halve the perceived latency figure in Sonar, but your CPU has twice the workload and you will need to increase buffer sizes much sooner to compensate as you increase the number of tracks/synths. For all intents the difference between 44.1 and 48 is imperceptible, and you should choose the appropriate rate for your release medium (CD, DVD etc) I think recording @ 24bit/44.1 is fairly common practice, and a good balance of resolution/headroom etc vs file size/performance, particularly when most final product will end up @ 16bit anyway.
post edited by tenfoot - 2016/04/04 03:43:05
Bruce. Sonar Platinum 2017-09, Studio One 3.5.3, Win 10 x64, Quad core i7, RME Fireface, Behringer X32 Producer, Behringer X32 Rack, Presonus Faderport, Lemure Software Controller (Android), Enttec DMXIS VST lighting controller, Xtempo POK.
|
AdamGrossmanLG
Max Output Level: -62 dBFS
- Total Posts : 1436
- Joined: 2014/07/13 03:40:34
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:04:05
(permalink)
tenfoot Doubling the sample rate will halve the perceived latency figure in Sonar, but your CPU has twice the workload and you will need to increase buffer sizes much sooner to compensate as you increase the number of tracks/synths. For all intents the difference between 44.1 and 48 is imperceptible, and you should choose the appropriate rate for your release medium (CD, DVD etc)
Thanks for responding tenfoot. Question, when you say "doubling the sample rate will halve the perceived latency figure in Sonar" - what do you mean exactly? What latency? like when I am singing into the mic? When playing an instrument? I'm not sure what this "latency" even is.
|
tenfoot
Max Output Level: -53.5 dBFS
- Total Posts : 2186
- Joined: 2015/01/22 18:12:07
- Location: Qld, Australia
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:12:43
(permalink)
Latency is the time it takes your system to process the audio - its round trip if you like. Generally when recording vocals and real world instruments you can monitor them directly using your interface software, so it is not such an issue. It really comes into play with soft synths and using real-time fx on inputs as you record (eg guitar fx inserts) If your asio buffers are too high, there will be a perceivable delay between playing a note on your vsti synth and hearing it. The setting in Sonar is under preferences/audio/driver settings, then click on the asio panel buttonn. Generally for tracking, the lower the better. Anything above 512 samples (around 11ms) and most people will be annoyed by the delay. Once you have finished tracking and you are mixing and adding fx, you can increase this setting to avoid any glitching that may start to happen as you increase processor load.
Bruce. Sonar Platinum 2017-09, Studio One 3.5.3, Win 10 x64, Quad core i7, RME Fireface, Behringer X32 Producer, Behringer X32 Rack, Presonus Faderport, Lemure Software Controller (Android), Enttec DMXIS VST lighting controller, Xtempo POK.
|
AdamGrossmanLG
Max Output Level: -62 dBFS
- Total Posts : 1436
- Joined: 2014/07/13 03:40:34
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:14:22
(permalink)
tenfoot Latency is the time it takes your system to process the audio - its round trip if you like. Generally when recording vocals and real world instruments you can monitor them directly using your interface software, so it is not such an issue. It really comes into play with soft synths and using real-time fx on inputs as you record (eg guitar fx inserts) If your asio buffers are too high, there will be a perceivable delay between playing a note on your vsti synth and hearing it. The setting in Sonar is under preferences/audio/driver settings, then click on the asio panel buttonn. Generally for tracking, the lower the better. Anything above 512 samples and you will definitely be annoyed by the delay. Once you have finished tracking and you are mixing and adding fx, you can increase this setting to avoid any glitching that may start to happen as you increase processor load. I
Thank you! So you are saying INCREASING the sample rate actually HELPS you with ratio when playing softsynths? I would think it would make the ratio WORSE, no? Because the system has to work harder?
|
tenfoot
Max Output Level: -53.5 dBFS
- Total Posts : 2186
- Joined: 2015/01/22 18:12:07
- Location: Qld, Australia
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:17:58
(permalink)
Not increasing the sample rate - increasing the buffer size under the asio setting described above will lower the system load.
Bruce. Sonar Platinum 2017-09, Studio One 3.5.3, Win 10 x64, Quad core i7, RME Fireface, Behringer X32 Producer, Behringer X32 Rack, Presonus Faderport, Lemure Software Controller (Android), Enttec DMXIS VST lighting controller, Xtempo POK.
|
tenfoot
Max Output Level: -53.5 dBFS
- Total Posts : 2186
- Joined: 2015/01/22 18:12:07
- Location: Qld, Australia
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:19:10
(permalink)
The sample setting in the asio panel\bufer size is not your sample rate. Leave your sampling rate setting @ 24bit (it is above the asio panel button)
Bruce. Sonar Platinum 2017-09, Studio One 3.5.3, Win 10 x64, Quad core i7, RME Fireface, Behringer X32 Producer, Behringer X32 Rack, Presonus Faderport, Lemure Software Controller (Android), Enttec DMXIS VST lighting controller, Xtempo POK.
|
AdamGrossmanLG
Max Output Level: -62 dBFS
- Total Posts : 1436
- Joined: 2014/07/13 03:40:34
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:19:45
(permalink)
tenfoot Not increasing the sample rate - increasing the buffer size under the asio setting described above will lower the system load.
but didnt you say: "Doubling the sample rate will halve the perceived latency figure in Sonar" - what does that mean exactly?
|
AdamGrossmanLG
Max Output Level: -62 dBFS
- Total Posts : 1436
- Joined: 2014/07/13 03:40:34
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:20:25
(permalink)
tenfoot The sample setting in the asio buffer is not your sample rate.
i know - i am talking sampling rate - in Sonar preferences.
|
musicjohnnie
Max Output Level: -86 dBFS
- Total Posts : 227
- Joined: 2009/09/20 18:38:43
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:23:32
(permalink)
Latency is, as I know it, the time it takes to produce a point ( note, tone, etc) and have that point taken, processed, and returned back to the ears to hear. Don't know if you have ever played a note on a keyboard and not had the sound played back at the same time you hit the note. Hard to overdub that way. I have a friend that produces everything at 92kh. He claims that sending music around needs to be higher to keep quality high enough for others to look at and use without degradation or loss. Hope this helps. Will help more if I can.
|
tenfoot
Max Output Level: -53.5 dBFS
- Total Posts : 2186
- Joined: 2015/01/22 18:12:07
- Location: Qld, Australia
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:24:09
(permalink)
As I said, increasing the sample rate setting also increases system load, so your asio buffer settings will need to be increased adding latency. The end result will be higher latency.
Bruce. Sonar Platinum 2017-09, Studio One 3.5.3, Win 10 x64, Quad core i7, RME Fireface, Behringer X32 Producer, Behringer X32 Rack, Presonus Faderport, Lemure Software Controller (Android), Enttec DMXIS VST lighting controller, Xtempo POK.
|
tenfoot
Max Output Level: -53.5 dBFS
- Total Posts : 2186
- Joined: 2015/01/22 18:12:07
- Location: Qld, Australia
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:25:02
(permalink)
Sorry - our replies are out of sync:)
Bruce. Sonar Platinum 2017-09, Studio One 3.5.3, Win 10 x64, Quad core i7, RME Fireface, Behringer X32 Producer, Behringer X32 Rack, Presonus Faderport, Lemure Software Controller (Android), Enttec DMXIS VST lighting controller, Xtempo POK.
|
AdamGrossmanLG
Max Output Level: -62 dBFS
- Total Posts : 1436
- Joined: 2014/07/13 03:40:34
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:30:32
(permalink)
oh no problem... ok so does the SAMPLING RATE (inside of Sonar) 44.1 vs 96 Khz have an effect on latency when playing softsynths?
|
tenfoot
Max Output Level: -53.5 dBFS
- Total Posts : 2186
- Joined: 2015/01/22 18:12:07
- Location: Qld, Australia
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:38:14
(permalink)
Doubling the sample rate halves the latency because the asio buffers fill twice as fast. With 1/2 the effective buffer time (there are twice as many samples for the same size buffer) than the system had at the lower sampling rate your processor will begin to feel the load much sooner, thus the need to increase your asio buffers, negating the perceived advantage of the higher sample rate.
Bruce. Sonar Platinum 2017-09, Studio One 3.5.3, Win 10 x64, Quad core i7, RME Fireface, Behringer X32 Producer, Behringer X32 Rack, Presonus Faderport, Lemure Software Controller (Android), Enttec DMXIS VST lighting controller, Xtempo POK.
|
AdamGrossmanLG
Max Output Level: -62 dBFS
- Total Posts : 1436
- Joined: 2014/07/13 03:40:34
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:41:44
(permalink)
tenfoot Doubling the sample rate halves the latency because the asio buffers fill twice as fast. With 1/2 the effective buffer time (there are twice as many samples for the same size buffer) than the system had at the lower sampling rate your processor will begin to feel the load much sooner, thus the need to increase your asio buffers, negating the perceived advantage of the higher sample rate.
wait, you are talking sample rate (as in 512 or 1024 in the ASIO settings), I am talking SAMPLING RATE inside of Sonar preferences (44.1, 48, 96, etc...). Which one reduces latency?
|
tenfoot
Max Output Level: -53.5 dBFS
- Total Posts : 2186
- Joined: 2015/01/22 18:12:07
- Location: Qld, Australia
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:47:39
(permalink)
Increasing asio buffers increases latency, but lowers processor load. Increasing the sample rate lowers latency (buffers fill faster) but increases system load. The end result (round trip latency) is interdependant on the two settings.
post edited by tenfoot - 2016/04/04 03:08:41
Bruce. Sonar Platinum 2017-09, Studio One 3.5.3, Win 10 x64, Quad core i7, RME Fireface, Behringer X32 Producer, Behringer X32 Rack, Presonus Faderport, Lemure Software Controller (Android), Enttec DMXIS VST lighting controller, Xtempo POK.
|
AdamGrossmanLG
Max Output Level: -62 dBFS
- Total Posts : 1436
- Joined: 2014/07/13 03:40:34
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:54:55
(permalink)
tenfoot Increasing asio buffers increases latency, but lowers processor load. Increasing the sample rate lowers latency (buffers fill faster) but increases system load. The end result (round trip latency) is interdependant on the two settings.
ohhh i see now - thank you for clearing that up for me.
|
tenfoot
Max Output Level: -53.5 dBFS
- Total Posts : 2186
- Joined: 2015/01/22 18:12:07
- Location: Qld, Australia
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 02:55:59
(permalink)
Bruce. Sonar Platinum 2017-09, Studio One 3.5.3, Win 10 x64, Quad core i7, RME Fireface, Behringer X32 Producer, Behringer X32 Rack, Presonus Faderport, Lemure Software Controller (Android), Enttec DMXIS VST lighting controller, Xtempo POK.
|
Sanderxpander
Max Output Level: -36.5 dBFS
- Total Posts : 3873
- Joined: 2013/09/30 10:08:24
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 03:49:35
(permalink)
SilverBlueMedallion
tenfoot Doubling the sample rate halves the latency because the asio buffers fill twice as fast. With 1/2 the effective buffer time (there are twice as many samples for the same size buffer) than the system had at the lower sampling rate your processor will begin to feel the load much sooner, thus the need to increase your asio buffers, negating the perceived advantage of the higher sample rate.
wait, you are talking sample rate (as in 512 or 1024 in the ASIO settings), I am talking SAMPLING RATE inside of Sonar preferences (44.1, 48, 96, etc...). Which one reduces latency?
If you want to use a higher samplerate to get a lower latency, that only works if your interface is already on or near its minimum buffer size and you have CPU power to spare. If you're not at minimum buffer size yet, adjusting that first will have a similar effect on latency and CPU power.
|
bitflipper
01100010 01101001 01110100 01100110 01101100 01101
- Total Posts : 26036
- Joined: 2006/09/17 11:23:23
- Location: Everett, WA USA
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 11:48:58
(permalink)
☄ Helpfulby mikekmac 2016/04/15 01:36:35
SilverBlueMedallion I have been reading that the HIGHER than sampling rate, the LOWER the latency? That sounds wrong! Wouldn't a higher sampling rate cause the hard drive and CPU to work even harder thus making latency even longer?
Yes, it sounds wrong because latency is more complicated than just how fast you shove bits into a buffer. Computers can't process audio data one sample at a time. They just don't work that way. Any time we bring data into a computer it has to be in chunks, whether we're talking about recording audio or reading data from a disk drive or a network adapter. Data may come in one byte at a time, but it gets stashed in a buffer until the buffer is full, and only then is the data actually processed. How long it takes to fill the buffer is therefore the main determiner of latency. You can therefore reduce latency by either making the buffer smaller or by filling it faster. At any given buffer size, latency can be reduced by sending it data more quickly, i.e. using a faster sample rate. However, you could also achieve the same effect by making the buffer smaller. Regardless of which method you use, the limiting factor is how quickly your CPU can process the data. At some point you will be feeding too much data too fast for the CPU to keep up. At that point you have no choice but to increase the buffer size, which negates the benefit of the higher sample rate, at least in terms of latency. But latency is more complicated than how fast you can get data into and out of the computer. There is also the overhead of what the computer's doing to that data. Many plugins necessarily introduce additional latency due to how they work - some must accumulate data within their own internal buffers because they work on chunks of data, too. Bottom line: don't increase sample rate in order to reduce latency. Use higher sample rates because you need the wider bandwidth, e.g. songs for dogs, dolphins or bats.
 All else is in doubt, so this is the truth I cling to. My Stuff
|
AdamGrossmanLG
Max Output Level: -62 dBFS
- Total Posts : 1436
- Joined: 2014/07/13 03:40:34
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 11:51:44
(permalink)
bitflipper
SilverBlueMedallion I have been reading that the HIGHER than sampling rate, the LOWER the latency? That sounds wrong! Wouldn't a higher sampling rate cause the hard drive and CPU to work even harder thus making latency even longer?
Yes, it sounds wrong because latency is more complicated than just how fast you shove bits into a buffer. Computers can't process audio data one sample at a time. They just don't work that way. Any time we bring data into a computer it has to be in chunks, whether we're talking about recording audio or reading data from a disk drive or a network adapter. Data may come in one byte at a time, but it gets stashed in a buffer until the buffer is full, and only then is the data actually processed. How long it takes to fill the buffer is therefore the main determiner of latency. You can therefore reduce latency by either making the buffer smaller or by filling it faster. At any given buffer size, latency can be reduced by sending it data more quickly, i.e. using a faster sample rate. However, you could also achieve the same effect by making the buffer smaller. Regardless of which method you use, the limiting factor is how quickly your CPU can process the data. At some point you will be feeding too much data too fast for the CPU to keep up. At that point you have no choice but to increase the buffer size, which negates the benefit of the higher sample rate, at least in terms of latency. But latency is more complicated than how fast you can get data into and out of the computer. There is also the overhead of what the computer's doing to that data. Many plugins necessarily introduce additional latency due to how they work - some must accumulate data within their own internal buffers because they work on chunks of data, too. Bottom line: don't increase sample rate in order to reduce latency. Use higher sample rates because you need the wider bandwidth, e.g. songs for dogs, dolphins or bats.
this explanation is ACE! thank you so much! I "get it" now! I am making relaxation music for dolphins so, I will be using sampling rate of 750 Khz.
|
Noel Borthwick [Cakewalk]
Cakewalk Staff
- Total Posts : 6475
- Joined: 2003/11/03 17:22:50
- Location: Boston, MA, USA
- Status: offline
Re: Sampling Rate / Latency Correlation
2016/04/04 12:08:23
(permalink)
☄ Helpfulby SilverBlueMedallion 2016/04/04 12:34:01
Its most easily understood by an example. Latency is always caused by buffering. Lets say you have a 512 sample buffer size. At a sample rate of 44.1KHz this translates to a latency of 11.6 msec. However assuming the same buffer size but a sample rate of 88.2 KHz this translates to half the latency which is 5.8 msec.
|