Attention:

Welcome to the old forum. While it is no longer updated, there is a wealth of information here that you may search and learn from.

To partake in the current forum discussion, please visit https://forums.presonus.com

Not getting sixteenth notes when recording

A Forum to Discuss NOTION

Re: Not getting sixteenth notes when recording

Postby Admin » Wed Aug 21, 2013 12:02 pm

The settings in the MIDI Record drop down need to be configured and experimented with to get the desired results. I find that sometimes I need to remove the capability to record tuplets to get dotted rhythms to record more accurately. Keep trying to find the settings that work best for the recording that you are doing.

Also, here is another point to consider:

Check your audio buffer size in the preferences dialog. Go to:

Preferences>Audio

Lower the Audio Buffer Size. If you are set to 256, you are hearing the sound at a quarter second of latency. That means that as you are recording into the system and hearing output from the system, there is some time between what you hear and what you play. This applies to the click as well, meaning that you could be hearing the click after the point in the score has passed. If you lower the buffer settings, you might get more desirable results.

-Admin
User avatar
Admin
Site Admin
 
Posts: 1006
Joined: Mon May 11, 2009 7:12 pm

Re: Not getting sixteenth notes when recording

Postby Surfwhammy » Wed Aug 21, 2013 8:26 pm

Admin wrote:Check your audio buffer size in the preferences dialog. Go to:

Preferences>Audio

Lower the Audio Buffer Size. If you are set to 256, you are hearing the sound at a quarter second of latency. That means that as you are recording into the system and hearing output from the system, there is some time between what you hear and what you play. This applies to the click as well, meaning that you could be hearing the click after the point in the score has passed. If you lower the buffer settings, you might get more desirable results.


I think this is incorrect--specifically the suggestion that 256 samples maps to a quarter second of latency. Everything else is fine, but I do not think that 1 sample maps to 1 millisecond of latency . . .

The Audio Buffer Size is a lookahead value, and it determines how many samples are preloaded, which affects latency but not in a 1 sample equals 1 millisecond mapping, where I think the mapping is more in the range of 25 to 50 samples equals 1 millisecond, depending primarily on whether getting a sample requires doing a hard drive read, which in the 64-bit universe should not happen very often, hence the ability to set the Audio Buffer Size higher, since all the samples are loaded into memory, and the lookahead buffer actually is doing memory reads and writes rather than hard disk I/O . . .

Among other things, this is the reason that when I am recording NOTION 4 generated audio as soundbites in DIgital Performer 8 (MOTU), I begin by clicking the "Record" button in Digital Performer 8 and letting it run for a few seconds, followed by stopping; rewinding; and then doing the actual recording. This preloads the lookahead buffers, and everything runs smoothly, but the key to the technique is to do it for just a few seconds, since if it is done too long, then what is in the buffer is wrong and defeats the purpose . . .

I think that what happens is that anytime during playback when there are a few idle cycles, NOTION 4 is coded to use the time to do a bit more lookahead, with the general goal being to keep the buffer filled at least enough always to have the next few notes loaded and ready to play. The full 1024 samples might not always be present, but there should be enough samples already loaded at any given time to map to smooth and timely playback, because if not, then it would be impossible to use NOTION 4 in a live setting where it is augmenting real instruments or providing instrumental accompaniment for a choir or singer . . .

This is what Ableton Live 9 reports when its Buffer size is set to 512 samples:

Image
Ableton Live 9 ~ Latency ~ Buffer Size

The guideline for Buffer Samples in Digital Performer 8 is that 128 samples maps to a latency of 3 to 6 milliseconds, depending on the processor speed and so forth, which is consistent with the latency calculated by Ableton Live 9, and I think the NOTION 4 Audio Buffer Size has a similar mapping . . .

THOUGHTS

I am very keenly attuned to echoes and delays, and if having the NOTION 4 Audio Buffer Size set to 1024 mapped to a latency time of 1 second, it would be intolerable . . .

As a general rule, I am fine with latency as high as 25 to 35 milliseconds, since it mostly sounds like very rapid digital delay, but when the latency is much higher I cannot sing or play electric guitar, because I need to hear what I am singing or playing on electric guitar essentially as it happens with nearly no latency (or ideally with no latency at all) . . .

256 milliseconds is a long time in this context--more so for lead guitar since it is possible to play several notes in that amount of time, and this is one of the reasons I usually record singing and electric guitar dry or if I need to hear effects, then I use real external digital effects pedals and echo units so that the only latency is the small amount that happens in Digital Performer 8 when there are no AU or VSTi effects units, which can be kept to a minimum via using the MOTU 828mk3 Hybrid external digital audio and MIDI interface . . .

It might be nice to be able to use AU and VSTi effects plug-ins in Digital Performer 8 when recording real lead guitar or singing, but the delay is too great, so I have Digital Performer 8 configured not to use AU and VSTi effects plug-ins for input monitoring purposes when recording real instruments and singing . . .

For reference, I did some experiments about a month or two ago, and I found that setting the Audio Buffer Size to 1024 maps to much smoother playback on the 2.8-GHz 8-core Mac Pro (early-2008) with 20GB of memory here in the sound isolation studio when running DIgital Performer 8 and NOTION 4 in 64-bit mode and using some of the MachFive 3 VSTi virtual instruments, since some of the MachFive 3 virtual instruments are highly sampled. . .

For example, the MachFive 3 Telematic electric guitar has 9,562 keygroups, and it takes approximately 45 seconds to load the keygroups from disk into memory during the NOTION 4 score setup step, which is separate from buffering the audio for playback and recording but is relevant in the sense of providing a clue to the quantity of samples that are likely to be required for playing a few seconds of electric guitar, which of course depends on the actual notes, articulations, and so forth . . .

It is possible that I do not understand the way the Audio Buffer Size value affects NOTION 4, but I do not think this is the case. Nevertheless, if my understanding is incorrect, please explain why and how to make it correct . . .

I think the actual latency is similar to the latency calculated for Ableton Live 9 based on its Buffer Size, which for NOTION 4 with an Audio Buffer Size of "1024 samples" puts the actual latency in the range of 25 to 50 milliseconds when it is running in 64-bit mode in Mac OS X 10.8.4 (Mountain Lion) on a reasonably fast Mac, where I use the qualifier "reasonably" because the Mac Pro here in the sound isolation studio is five years-old and there are considerably faster Apple computers today, although the Mac Pro here in the sound isolation studio continues to be what I consider to be "peppy" . . .

And I think this is easily verified by a simple experiment, where you set the tempo to 60; use the mouse to input several measures of quarter notes for a snare drum in 4/4 time, using the NOTION 4 Bundled Drum Set; position the cursor at the first quarter note in the series; and then press "Play" . . .

If the Audio Buffer Size set to "1024 Samples" maps to a 1-second delay then you will not hear the first quarter note until 1 second later, which is a very long time and is the rate at which you will hear the quarter notes, since at this tempo there are 60 beats per minute with each quarter note getting one beat . . .

When I do this experiment, the delay between clicking on the "Play" button and hearing the first quarter note is nearly imperceptible. The delay is well within the tiny range of what I expect to hear when I am playing a real snare drum . . .

However, it is not instantaneous, and I can tap the space bar twice very rapidly and keep the cursor in the same location between two quarter notes, as shown in the following screen capture, so that the snare drum is not heard and the cursor stays in the same location, and I also can tap the space bar four times and keep the cursor pegged to the same in-between location, which is probably due to always having a bit of fun playing "finger drums" on tables, which is one of the finger exercises I do frequently . . .

Image
NOTION 4 ~ Cursor between notes

Regarding playing "finger drums", I do drum rolls with the fingers and thumb of each hand, usually in sequence from outside to inside (or little finger to thumb), and I alternate hands to do very rapid "finger drum" rolls . . .

[NOTE: For the genres I like, this works very nicely, and my strategy for playing piano is to play it as if each key is a separate drum or cymbal and my fingers are drumksticks, and so long as I compose songs in one of the keys signatures where all the white keys sound "good", if I only play white keys, then the worst that happens is I change modes at odd times (Ionian, Dorian, Phrygian, Lydian, Mixolydian, Aeolian, Locrian) or whatever. And if I want to make the piano notes a bit dissonant, then I switch to the black keys for a while . . . ]

I did a few experiments several years ago and determined that it is trivial to play 20 notes in 1 second on a standard piano keyboard this way, which maps to 1 note every 50 milliseconds, although the notes are not so melodic, but it is consistent with Arnold Schoenberg's Twelve-tone Technique . . .

I did this because I was interested in the Duallist Single-Foot Double Pedal and needed to know whether I needed two of them to be able to play 70 beats on one or perhaps two kick drums in 1 second, where the current perspective is that for me it requires two of them, one for each kick drum, although at present I use a Duallist Single-Foot Double Pedal only for my right foot but use a single-foot single pedal for the left foot, which is fine until I can afford to get the second Duallist Single-Foot Double Pedal. I am not getting 70 beats per second, but I think it is in the range of 35 to 50 beats per second, which is pretty good . . .

SUMMARY

I am quite confident that setting the Audio Buffer Size to "1024 Samples" does not map to a latency time of 1 second . . .

Lots of FUN! :ugeek:
Last edited by Surfwhammy on Thu Aug 22, 2013 11:43 am, edited 1 time in total.
The Surf Whammys

Sinkhorn's Dilemma: Every paradox has at least one non-trivial solution!
User avatar
Surfwhammy
 
Posts: 1137
Joined: Thu Oct 14, 2010 4:45 am

Re: Not getting sixteenth notes when recording

Postby Admin » Thu Aug 22, 2013 11:05 am

Sorry for that misinformation. You are correct about the latency. For some reason I thought it was milliseconds instead of samples.

We operate at 44.1kHz, so 256 is next to nothing on the scale. But, I do know that 1024 and even 256 are noticeable and rather distracting. For something as necessarily accurate as MIDI Record, I still think that turning the Audio Buffer down will help the accuracy of the MIDI Recording.

-Admin
User avatar
Admin
Site Admin
 
Posts: 1006
Joined: Mon May 11, 2009 7:12 pm

Re: Not getting sixteenth notes when recording

Postby Surfwhammy » Thu Aug 22, 2013 11:26 am

Admin wrote:Sorry for that misinformation. You are correct about the latency. For some reason I thought it was milliseconds instead of samples.

We operate at 44.1kHz, so 256 is next to nothing on the scale. But, I do know that 1024 and even 256 are noticeable and rather distracting. For something as necessarily accurate as MIDI Record, I still think that turning the Audio Buffer down will help the accuracy of the MIDI Recording.

-Admin


No problem! Glad to help! :)

"256 Samples" was an excellent value for NOTION 3 on the Mac Pro here in the sound isolation studio, but after experimenting with 64-bit NOTION 4 for several months and noticing that some of the MachFive 3 (MOTU) virtual instruments were a bit choppy at times, I increased the Audio Buffer Size to "512 Samples", which helped to smooth the MachFive 3 virtual instruments. And then I increased it to "1024 Samples", which is working nicely but might be a bit high for recording real-time MIDI as accurately as possible, where for reference recording real-time MIDI is a new activity for me, so I continue to do experiments to determine the optimal way to do it . . .

As best as I can determine, with the increased memory that comes with running in 64-bit mode, more of the samples are loaded into memory from the hard drive when a NOTION 4 score is opened, which in turn makes having a larger Audio Buffer Size acceptable, if not for some purposes desirable . . .

On the other hand, having a smaller footprint when recording real-time MIDI makes sense, where the general idea as you correctly noted is to create the fastest configuration for getting real-time MIDI recorded as accurately as possible, which is fabulous . . .

Fabulous!
:)
The Surf Whammys

Sinkhorn's Dilemma: Every paradox has at least one non-trivial solution!
User avatar
Surfwhammy
 
Posts: 1137
Joined: Thu Oct 14, 2010 4:45 am

Previous

Return to NOTION

Who is online

Users browsing this forum: No registered users and 17 guests


cron