slow attack and weird reverb

[marclabru]

I'm trying to put a free reed clarinet in my CODM organ. But the attack on some of the notes are much slower than with the original organ loaded. Also, the reverb sounds not good (synthetic) Are these problems because of limitations with CODM format? or am I doing something wrong? I read about attack samples, are there maybe seperate attack samples for this samplesett I haven't found? and where do you put those in the CODM file?

[mdyde]

Hello marclabru,

I think you probably have the Rank attributes that specify the output pitch and/or sample pitch and/or sample note number mapping set incorrectly, resulting in the samples being pitched down (e.g. by an octave). That will make their attacks and releases sound unnaturally long.

E.g. Check:

- Rank.Samples_MIDINoteNumIncrementFromPipesToSamples
- Rank.Samples_RankBasePitch64ftHarmNumIfAssumedTunedToConcertPitch [or the pitch values stored in tthe sample files, if applicable instead]
- Rank.Pitch_RankBaseOutputPitch64ftHarmonicNum

[ldeutsch]

First, I totally agree with Martin's reply.

Second, there is indeed an issue here - even if you are not re-pitching the samples. I often assemble CODM organs from multiple sample sets. Sometimes I discover pipes (or even entire ranks) where the original sampler has perhaps erred on starting the attack samples too early, resulting in a slower-than-desired attack. There is currently no way to compensate for this within the CODM - even though there IS a way too compensate for releases that are too long (truncation.)

Perhaps we should add a mechanism of some sort for this to the CODM. Like truncation, but in reverse, this would allow the user to define the length of the attach by starting the wave shape at some time other than zero, while introducing a ramp so that the pipe does not produce a click be starting at an amplitude larger than zero.

The only real alternative to this involves editing the wave files themselves - which is perhaps beyond the capability of most users and, in any case, much more work!

Martin, if you want to discuss possible algorithms for this with me, send me a private message.

Les

[mdyde]

Hello Les,

Thanks very much for the suggestion. I've logged it as an enhancement request for long-term refernece, but I don't think I'd view it is being a high priority. I think my view would be that it would be better (and not too difficult) for people to could edit the sample files directly in such cases.

[marclabru]

Thank you both for your reply.
Martin, I saw I had to change the rank base pitch, unfortunately it didn't change much. So it was good to hear that slow speach can be an issue as Les said.
I still encounter som metallic unnatural sound in the end of the reverb, it's not there in the originals.
Should CODM produce the same result, or canthere be differences due to the limitations?

[mdyde]

Thanks, marclabru.

For good measure, with the organ loaded, please first go to the "Organ settings | Organ preferences | Wind supply model" screen tab and tick the option to disable the wind supply model (since a misconfigured wind supply model in the CODM ODF can cause the pipework to sound different).

Also, make sure that you haven't added any impulse response reverb on Hauptwerk's mixer (or externally to Hauptwerk), since reverb will affect how pipes sound, and adding reverb to samples that already contain reverb can sometimes sound unnatural.

Assuming those things don't solve it:

If, in Hauptwerk, you play a single pipe for which a corresponding sample file exists, hold it for a while, and then release it (and assuming that the release marker 'cue point' is positioned appropriately in the .wav e file), then it should sound essentially the same as playing the corresponding sample file in an audio editor. If it doesn't, make absolutely sure that the expected samples are playing on the intended keys in Hauptwerk, and at the same pitches.

For example, find one of the sample files that has a unique and distinctive sound somewhere in it, compared to the others (e.g. a speech artefact, or some noise in it) and check that in Hauptwerk it's playing on the expected key, and at the same pitch as the original sample file. If not, your CODM ODF Rank pitch attributes must be misconfigured (and/or the pitch stored in the sample file must be incorrect).

[marclabru]

I found out it's just a matter of finding the right release samples. (choose the right 2 out of 3).
It would be nice if the CODM could have the possibility for 3 release samples in the future, since a lot of sample sets have 3

[mdyde]

Hello marclabru,

Thanks for the update, and glad to hear you managed to resolve it.

It would be nice if the CODM could have the possibility for 3 release samples in the future, since a lot of sample sets have 3

Actually, it already does. Have a look at the documentation for the following Rank table attributes:

Samples_RelSampleFileFolderIfRelSamplesStoredInSeparateFiles
Samples_ShortNoteReleaseSampleFileFolderIfAny
Samples_MaxNoteTimeMsForShortNoteReleaseSamplesIfAny
Samples_MediumNoteReleaseSampleFileFolderIfAny
Samples_MaxNoteTimeMsForMediumNoteReleaseSamplesIfAny

[marclabru]

Thanks, I meant 3 release files in addition to the main release file folder So actually four. But it seems only Friesach has it

[mdyde]

Ah. Thanks for the clarification.

[tf11972]

If you record also tremmed stops you have actually four releases per pipe. When you press the keys manually during sampling, it‘s not possible to produce samples which lie exactly between portato and staccato. Only a machine could do that.

[ldeutsch]

Actually, most of the time when you have tremmed and non-tremmed versions of a rank using the CODM, they will be treated as two separate ranks - so the number of releases "per rank" does not increase even though the number of releases "per stop" doubles!

Also, I have been involved with several pipe organ sampling projects where we have actually used a MIDI player attached to the pipe organ to play the notes to be sampled. If this is available, it allows one to sample a great many pipes multiple times - perhaps in the middle of the night when there is less noise in the building.

Les

[tf11972]

The time range to press a key is rather small, a short press has about 150 ms, a medium about 500 ms. All above are long releases. In which steps should a MIDI-player press the keys? 100 ms or even 50? And does the human ear hear the differences or can HW and the computer handle the lots of releases when you only hear over headphones?
Theory is always nice, the practical implementation difficult.