NOTION Music Forum

I am trying to setup Notion to slave to Sonar x2 and I can get the playback working. However all my projects are recorded using 48khz and it seems that Notion will only use 44.1khz. There is an option to tune the audio playback but it will not quite go far enough to get to the correct pitch.

I know I can continue to work around this with imported audio and such but was hoping not to have to go that way.
Or I could transpose the entire score down a half step and tune from there but that just makes my skin crawl LOL.

Is anyone else using a sample rate higher than 44.1khz? If so would it benefit you to have Notion be able to tune itself for higher sample rate playback?

If there is sufficient interest it should not be too difficult to expand the tunable range of the audio to accommodate.

If this is something that would interest you drop a note and perhaps we could submit a feature request.


For now looks like I'm stuck doing it the old way or converting my projects before using Notion 4.


Tom

tomr wrote:I am trying to setup Notion to slave to Sonar x2 and I can get the playback working. However all my projects are recorded using 48khz and it seems that Notion will only use 44.1khz.

You asked several intriguing questions, but instead of focusing on the specific questions, I think it will be more helpful to provide a bit of advice and some related information . . .

REFERENCE TUNING PITCH

The standard "Concert A" reference tuning pitch is 440-Hz, but over the past few decades some folks have decided that other reference tuning pitches are better, and there is merit to their perspectives, since for example the reference tuning pitch(s) during the times when much of what one might call "Classical" and "Baroque" music was composed and played were different, as were the instruments, where for example the strings for violins, violas, cellos, and contrabasses, as well as guitars, pianos, harpsichords, and so forth were made of different material, hence what one might call "purists" strive to do everything as consistently and accurately as possible based on the factual realities during those times, which is fine . . .

By default NOTION 3 used a non-standard reference tuning pitch, so it had to be set explicitly to 440-Hz if one wanted to use 440-Hz, which is the way I do things here in the sound isolation studio, and NOTION supports this by making it possible to change the reference tuning pitch, all of which is fine . . .

However, there is more to it than simply changing the reference tuning pitch, because NOTION uses sampled sounds to generate audio . . .

It might be the case that the NOTION bundled sampled sounds from the London Symphony Orchestra were recorded with the various instruments tuned to a different reference tuning pitch, in which case it makes sense for the default reference tuning pitch to match the NOTION bundled sounds, but nearly everything else continues to use 440-Hz as "Concert A" . . .

What happens behind the scenes is that the reference tuning pitch is applied to sampled sounds and as such becomes a multiplier or whatever one wants to call it, and if a sampled sound library was recorded using 440-Hz as the reference tuning pitch but the reference tuning pitch in NOTION is set to 455-Hz, then every sampled note will need to be adjusted accordingly, because otherwise it will be out of tune or detuned . . .

The problem is that it is not simply a matter of doing a bit of simple multiplication in every instance, and one of the consequences is that what would be pristine samples which did not need to be mathematically adjusted then become different . . .

As an example, if you record a note played by a violin tuned to 440-Hz as "Concert A" but later decide that 455-Hz is better, everything needs to be adjusted, and while the simplistic adjustment is to multiply by 455/440 (which is approximately 1.03409090909091, common sense suggests that if the note was sampled for 1 second, then what happens to the sample with respect to texture, duration, harmonics, overtones, and so forth and so on . . .

At the extreme, what happens is similar to the way the voices of the Alvin and the Chipmonks are done, which is fine for novelty songs but not so stellar for accurate recorded instrument sounds . . .

So, without going into more detail on all the algorithms, my perspective is that it is useful to have a limited range of adjustments to the defined reference tuning pitch, but anything beyond what NOTION provides has consequences that in nearly every instance will be undesired . . .

This is easier to understand when you consider that the aforementioned note was played by the musician using vibrato or perhaps was played through an amplifier using a tremolo effects pedal, and what happens is that while the pitch and tone of the note might sound reasonably accurate but a tiny bit higher or lower depending on the multiplier, the vibrato or tremolo speed will be different, which has consequences, because especially with vibrato, what happens is that by applying a mathematical algorithm essentially to compute an artificial note, the vibrato performed by the skilled musician is altered, which makes it a destructive process, where using a computer you arbitrarily decide that all the years of training, practicing, and playing done by the violinist toward the goal of becoming a master player essentially is meaningless or whatever, because when you do the mathematical algorithm the computer overrides the judgment, skill, and mastery of the violinist, and I suggest that this is not the brightest strategy in the grand scheme of everything . . .

SAMPLE RATE CONVERSION

The smarter solution to the problem you described is to convert the already recorded audio files to 44.1-kHz at 16-bits, and most Digital Audio Workstation (DAW) applications can do this, where it appears that Sonar (Cakewalk) has this ability . . .

Steinberg has an industry standard product, WaveLab 7, which is used for mastering, sample rate conversions, and other advanced audio file editing and enhancing, so it is a possibility depending on your budget, and for reference the MSRP is $499 (US) . . .

I did an experiment with Digital Performer 7.24 (MOTU) where I converted the audio files for a song from (a) 44.1-kHz and 16-bits to (b) 48-kHz at 16-bits, which is converting upward rather than downward, and the song sounds the same to me, where specifically it is the same key and "Concert A" continues to be 440-Hz after the sample rate conversion, and the tempo is the same, as well . . .

There were several hundred audio files, and it took approximately 1 hour on the 2.8-GHz 8-core Mac Pro (20GB memory) to do the sample rate conversions, which on average used 50 percent of the 8-cores, which is clue to how much computing is required, and the algorithms used by MOTU are high-quality, as will be the algorithms used by Steinberg in WaveLab 7 . . .

From the various information I found about Sonar, I think the Sonar algorithms also are high-quality, and there might be a way to do all the audio files for a song or project as a set rather than one file at a time manually, which is one of the options in Digital Performer 7.24 and makes it vastly easier, although it takes a while . . .

The best way to do the sample rate conversion is to make a copy of the songs or projects, including all the associated audio files and so forth, and then to do the sample rate conversion on the copy rather than on the original audio files, since you want to preserve the original audio files . . .

And for reference, there is a 30-day trial version of WaveLab7, which might be enough time to do the sample rate conversions, although if you do some experiments with Sonar and are pleased with the results, then you can to the sample rate conversion with Sonar, which you already have . . .

Going from 48-kHz to 44.1-kHz will result in losing some information, but the reality is that most people cannot tell the difference, and by the time a song is mastered and appears in the iTunes Store or is played on FM or satellite radio, it is different, so my perspective is that the lost information is not significant in a practical way . . .

And if you listen to songs in MP3 format, I think it is easy to understand the practical reality, because high-quality MP3 songs sound very good, but it is a compressed format and in some instances will have a variable bit rate (VBR), which is an acoustic physics technique that improves the overall quality by varying the fidelity depending on the particular sounds and so forth at any given time, where some sections or time slices of a song can be accurately reproduced with fewer bits of information while other sections or time slices requires more bits of information . . .

Lots of FUN!

Thanks for the reply. I understand the nature of sample sets and the like but work with me here.

Notion is using 44.1khz sample sets. I drive Notion from a 48khz daw and it naturally produces notes that are a little more that 1/2 step sharp. Because it thinks it is playing at 44.1khz.

Stay with me. If I drive Notion at 48Khz and reduce its audio engine frequency so that its sample sets are playing back at the speed and frequency that were recorded at 44.1. Everything will sound correct.

The existing controls are sufficient to do this if the constraint placed on the tuning slider was increased just a bit.

Edit: I just did the calculation. Currently Notion allows the clock speed to be reduced by %5.611 of the default A442. If it allowed adjustment to at least %8.125 then the samples recorded at A442 would playback at A440 if the rewire clock driving notion was running at 48khz. All they would need to do is add a check box in the audio where the enable rewire box is that would allow you to set the speed of the in coming rewire clock and do this adjustment for us in the background. No need to change the samples at all.

The intonation curve you are speaking about is a known quantity and can be modified on the way out just as it is currently. You are right in that if the frequency varies too much it will sound bad.

But if Notion was to decide they wanted to allow say 48khz and 96khz etc the best way would be to prepare samples for each frequency to do it cleanly. Making the sample sets much larger. (what fun that would be to download)

For instance. I was able to use the transpose function in Notion to lower the score 1/2 step and then use the existing tuning slider to reduce it he rest of the way and the playback sounds correct. This method makes the score look funny but unless I try to exceed the range of the instrument it should work but is less than desirable.

Yes I can convert the audio files however there are lots for each song and they are each in some state of partial completion. So the conversion would be a pain. As the best method would be to create a new project using 44.1 and import each file into the project. They will get up or down sampled at that time.

tomr wrote:I can convert the audio files however there are lots for each song and they are each in some state of partial completion. So the conversion would be a pain. As the best method would be to create a new project using 44.1 and import each file into the project. They will get up or down sampled at that time.

I think this will be the best strategy, and it will map to the best sounding songs, which is fabulous . . .

Fabulous!

Thanks for your interest Surfwammy. I may consider starting any new projects at 44.1 but I will not be converting existing ones.

I will submit a feature request and see what happens. Obviously from the lack of others responding they are satisfied with the 44.1 playback so not much interest in going beyond that here on the forum.

Cheers