< Back to IRCAM Forum

Filtering intervals


I have a sdif converted into a voice but now I would like to be able of filtering some intervals. Is it possible? I really appreciate any clue.

Thanks in advance

Hi Xoanxil,

Can you be more precise? A Minimum working Example would be great.

Hi Karim,

I have not my computer with me right now, but I am doing some basic trials with this process:

  • Creating sdif files from audio recordings using Spear

  • Opening them in OM using Sdif and convert them into notation using Sdif->chord-seq

  • Quantify and convert them into voice

  • I get some interesting musical material but, obviously, with a lot of intervals that, in this case I would like to avoid, as minor seconds.

My question is how could I filter some intervals for not be included in the resulting voice or chord-seq.

I really appreciate any clue.


Hi Karim,

what is Spear ? is it exist on OSX ?



Hi, Again,

Yes it is possible. I can send you an example starting from a voice. However, can you elaborate.:

  1. are there any Chords, or just a melodic material. This is important since in the last case this should be really simple. For chords it is more complicated

  2. what should we substitute these “filtered” intervals with ? rests? or another interval? This is also an interesting option.


Hi smalllotus,

Here is Spear: http://www.klingbeil.com/spear/
it is an Sinusoidal Partial Editing Analysis and Resynthesis software that resembles our AudioSculpt.


1 Like

Thank you very much Karim… :slight_smile:

Just awesome !!!


Now I am doing some tests so I am trying Chord-seq for chords and Voice for melodic. I was thinking just delete one of the two notes of the interval, but replace with other intervals is an interesting point.

As soon as I get home I can send you an example of the working flow I am using.


I have here a patch that will remove not wanted intervals in the chord.
In my example , unissons and minor seconds but it could be any.
tell me if this is what you are looking for.

Take care
xoanxil 2.omp (38.4 KB)

¡Thank you very much Karim! Any clue about how to address this process with chords?

Here you can find a very basic patch I am using after clean audio recordings using Spears and generating a sdif file.

But I get very complex chords and I would like to have more control filtering intervalic content.

Dear Xoanxil,

Well i believe the patch i’ve sent deals with chords intervals(?) Or maybe i didn’t really get your request?

Can you be more precise and send just the voice as a patch?


I am trying to upload a file but it said “Sorry, new users can not upload attachment”.

So, I have been doing some more tests. I am using your patch here:

From the upper Voice to the botton transformation it should remove second minor intervals but it looks not to be working. I get a C sharp and D at the end of second bar.

I am openmusic newbie so maybe I am missing something, sorry.

Thank you very much for your patience Karim.


My bad. didn’t figure this out. Here is a patch that should fix it.

xoanxil 3.omp (38.6 KB)

Tell me if it works for you.

By the way, I am very sorry for the “Sorry, new users can not upload attachment”. It is really stupid. I believe it is because who did configure this, must be some kind of paranoid. I will report this so they can fix it. I am sure you are no robot! :slight_smile:

Thank you very much Karim. Now it is working! I am having some problems with the way I am creating harmonic progresions from .sdif material but I guess it is my fault.

I am using align-chords and when I try to filter this Voice material into a new voice with the intervalic filter I still get semitones but maybe my aproach using align-chords is not a good one.

Any idea how could I manage this?

Hi Xoanxil,

What i suggest (what i do usually), first is not to rely on the quantification preferences.
I would use instead as->om. It’s fourth input is delta time for chords (alike align-chords). Then you can also choose your approximation, and number of polyphony.
After building your chord-seq, i will use omquantify explicitly, and choose the correct quantification i need AFTER, setting in om’s preferences (quantification) tab the Delta Chords to 0 ms.
Will send you an example on how i deal with this. It might not be the perfect solution, but it is my preferred method.

…By the way, when you say “i still get semitones”, this means you are looking to transcribe exactly some music from audio to midi. Am i correct? If it is the case note that this is not an automatic transcription, and you will not have exactly a perfect transcription. I think that you need first to “prepare” your analysis in Spear, cleaning your source, etc.

Thanks Karim. I am working with fieldrecordings. I am not looking for getting a perfect transcription but using some of this raw material in order to get something conected with it to work on it later composing for acoustic instruments.

I am using in spears “below threshold” and “below durations” to reduce information. Would you suggest another aproach in spear before export as .sdif?


I have noticed other problem. Even when I have a silence in my spear file OM doesn´t convert it into rest. It is more like if it extends the last note.

I mean, it made it nice from as->om to chord-seq but when I pass it to a voice it transform everything too much. What shoud I do for Voice doent change chrod-seq durations.

Any idea? Thanks!

Dear Xoanxil,

Some notes on quantification:

First you must know that when you connect a Chord-Seq to a Voice, you are also quantifying the duration. You will therefore find all default values in the quantification tab in om preferences.

  1. In the above figure, (1) it’s what we have done. You will notice that all notes are “wrong” shifted, because: a) we have set the delta value to 0, so all durations are individual and not collected in a delta time. You will also notice that we have “lost” 72 notes in the quantification, and therefore, the notes are shifted compared to the durations.

  2. Here we used the align-chords and set delta time at 100. This is equivalent to the setting in the preferences, so all durations falling below 100 ms will be concatenated into chords. That’s what we want so to avoid "lost"notes.

  3. Now the issue of sustained notes, which is very annoying of course comes from the fact that the last note of each group here, has a very short duration (10ms for the last note of the first group as we can see in the output of true-duration in the listener). Here unfortunately, we cannot do but to edit it “by hand”. The reason of this all is that gracenotes are not yet implemented in OM. and these notes are even after-grace notes. However, there is an ongoing project to do so. So for the time being you should edit these by hand as i do (see voice 5 in the patch. I edited the first 3 groups).

quanti.omp (226.6 KB)
PS: Concerning Spear, i am sorry i don’t use it. So my help here is nil

Thank you very much Karim for you patience and explanations. This is really helpful and very enlightening!

I gues that now I have enough information to start to experiment and look for the best setting for each material.