SynthFont.com

Admin wrote:I got them alright!
Try this version:
https://dl.dropboxusercontent.com/u/173 ... 2020I1.rar

This version plays using GM.dls but sounds out of tune as ray says!

It doesn't sound out of tune when using the MIDI OUT (Microsoft GS Wavetable Synth) which uses the same GM.dls!

EDIT: When exiting SynthFont from I or I1 I get an error report, not sure if you are receiving those or not.

I haven't got any new error reports. Why don't you try running THIS version with DWIN32?
https://dl.dropboxusercontent.com/u/173 ... 2020I2.rar

Admin wrote:I haven't got any new error reports. Why don't you try running THIS version with DWIN32?
https://dl.dropboxusercontent.com/u/173 ... 2020I2.rar

Nothing out of the ordinary in the output.

Yes, I get the same when I have set gm.dls as the default soundfont. There is something fishy in the dls read routine.

Admin wrote:Yes, I get the same when I have set gm.dls as the default soundfont. There is something fishy in the dls read routine.

Oh perhaps we're getting our wires crossed, I was testing the exceptions thrown on exit, this has nothing to do with the DLS issue, was that debug version for testing the pitch problem or the exit exception? (Which appear to be two completely different problems.)

Indeed, the exception had nothing to do with DLS. In my case it was due to me renaming the VST folder to something else. Got that sorted out now. Your issue may be something similar, but I doubt it. That would be a real coincidence!

Anyway, I have found the bug causing wrong pitch for some notes in a DLS file. It is a DLS specific problem and it has been with me since the very first years! How strange that it has gone unnoticed for so long. It is in Synthfont1, Viena, SyFonOne, VSTSynthFont, but no-one has noticed it before.

Here is version J:
https://dl.dropboxusercontent.com/u/173 ... t2020J.rar

Admin wrote:Indeed, the exception had nothing to do with DLS. In my case it was due to me renaming the VST folder to something else. Got that sorted out now. Your issue may be something similar, but I doubt it. That would be a real coincidence!

Anyway, I have found the bug causing wrong pitch for some notes in a DLS file. It is a DLS specific problem and it has been with me since the very first years! How strange that it has gone unnoticed for so long. It is in Synthfont1, Viena, SyFonOne, VSTSynthFont, but no-one has noticed it before.

Here is version J:
https://dl.dropboxusercontent.com/u/173 ... t2020J.rar

This version appears to fix both issues. Huzzah! =)

Two bugs down, now for quick vids of the other bugs I reported earlier:

- MIDI Files with effects may have inconsistent pitch with lower sample rates

- Cut previous note short/off and legato settings under overlapped note handling takes the "Do Nothing" behavior (Since 2.0.1.2R)

- Audio output stalls for a period of time in MIDI Input mode if several All-Notes-Off messages are sent (Since 2.0.2.0E)


As well, I've discovered another bug originating all the way from the second test build:

- Re-sizing the program horizontally causes the stop marker to be incorrectly placed on the mini-pianoroll (Since 2.0.1.2B)

Here's version "J1":
https://dl.dropboxusercontent.com/u/173 ... 2020J1.rar

The only real bug fix is the "Re-sizing the program horizontally causes the stop marker to be incorrectly placed on the mini-pianoroll" issue. The fix isn't very elegant, but it works. (It actually should have worked already as there is code for responding to a resized window, but for some really odd reason this didn't work, so I had to resort to a hack).

Now about the other issues:
- "MIDI Files with effects may have inconsistent pitch with lower sample rates" - Well, 1) the particular file you have used as example doesn't actually contain any "effects" or controllers of value. It is only notes, one after another. 2) the standard pitch shifting algorithm isn't good enough for downshifting to 8000 Hz. I'm not surprised at all. The standard method is so simple (and fast) that it easily introduced artifacts in the sound. Therefore I have the Perfect Pitch method, which is slow but almost perfect. Give it a try. [The "PSM" button in the tracks header, to the right]

- "Cut previous note short/off and legato settings under overlapped note handling takes the "Do Nothing" behavior" - I don't agree. See this document for a description of how its works - and that it works: https://dl.dropboxusercontent.com/u/173 ... _notes.pdf. Your video is a little bit odd. It sounds to me like you have used a different SoundFont for the legato case.

- "Audio output stalls for a period of time in MIDI Input mode if several All-Notes-Off messages are sent" - can't reproduce. Here is a video: https://dl.dropboxusercontent.com/u/173 ... _forth.mp4 Maybe it just takes a lot of time for SynthFont2 to search for a SoundFont that contains a preset for the specified bank on your computer. Without further analysis I can't tell. How does it work if you play another MIDI tune, one that uses presets in bank 0 (and 128) only?

The new 2.0.2.0J1 fails at playing Springtime Sundays without cutting the notes before their end (Note events are being ignored).
Affected MIDI: https://mega.co.nz/#!ZFEz0BoQ!6BO9hb8QC ... YijP6cvhXQ

EDIT:
This problem only occurs on 2.0.2.0, but not on 2.0.1.1 (Where there's the right channel bug.)