Help with mellotron (channel 221)

[SuNo-san]

I could not make it work (yet) - any advice ?

Checklist:

+ loopsynth v2 
+ v1.xxx.118 
+ mellotron directory and many subdirs
+ .wav in the directories
+ mellotron directory on the SD card (checked twice)
+ 16-bit or less samples (VLC says so) - by the way, I am using https://sonicbloom.net/en/free-sb-mellotron-samples/ more precisely the Flute MK2 [attached]
+ Arturia Keystep 32, output midi channel 1, midi cable OK (tested elsewhere) connected to the DIN-TRS adapter (tried two of them, type A)
+ load channel 221 - seems OK (steady RDY yellow LED)
+ some signal seems to be coming in (kind of a small blue light led SIG is blinking)
+ volume B4 at high level (tested on channel 1)

-- no sound 

Of course, this might be difficult to debug... but anyone can see what could be missing ? E.g. Midi Channel number ; should I "tune" something on the Loopsynth to get the Keystep 32 recognized further? I used it to define a chord progression, seems to work fine.

Gilbert

[mario]

Hi Gilbert,
Thank you for detailed info, and the rar file.
I see that subdirs are all named with extra "-3" why is that?
They should simply contain a note (and possibly octave) e.g. "a#2".
The wav files are good to be named as they are.
Fingers crossed it will start working!
Cheers,
Mario

[SuNo-san]

Blindness... the naming came from bulk processing of the original filenames...

Now it works, however only with one note at a time ; playing two or three creates some artefacts : some noisy interferences that I don't know how to describe - I could record a sample if you can't reproduce it. attached is the correctly named directory.

Also, contrarily to the doc, I can play more than 3 notes at a time (I believe 10? Have no more fingers).

G.

[mario]

Now it works, however only with one note at a time ; playing two or three creates some artefacts

The reason is that there is no automatic level normalization, or any control over their mixing volume, data is simply added together. If the samples are loud enough, even mixing two of them will likely overshoot the headroom. This requires better preparing the samples. For example, normalizing by peak to -10dB will make sure that no sample is beyond 33% of the dynamic range, mixing 3 voices will then never cause clipping. Normalizing by RMS / loudness perhaps makes more sense, but I'm not sure how to choose the levels, may require experimenting to find out where will they sound balanced. Or maybe just adjust the volume of each sample by 50-60%, chance is that the clipping will be gone.

Also, contrarily to the doc, I can play more than 3 notes at a time (I believe 10? Have no more fingers).

If you play more than 3, the newer voices will be replacing older, there is a limit to how many files on the SD card can be open and read from at the same time. It is probably hard to notice what happens now that it all sounds distorted.

[SuNo-san]

Thanks Mario. I will try that on the samples I want to load - I used these as a test of the functionality.

G.