dv wrote:

New kid on the block of audio-to-MIDI translation…

Now I get it: why all the negativity, supposedly about a DAW.  I just completely disregarded your previous post.  Man, this was quite totally unintentional! 

I just edited my own; no way now does it contradict or supplant yours.

No need now to further bash DAW usage, let’s just see if people like you and I positively can get along somehow  —with even different views, of course about a DAW.  (Smile.)

I didn't intend to insult or anything, I apologize for the negativity. We can be friend and I'd be happy to. Just that I won't use DAWs because I don't need them (but I do understand people who do). That asides, I don't get why one would want this paid new kid on the block. Well maybe it's more accurate or faster or better or whatever than the free ones, but you did not say that :-) so I don't see the point of paying

New kid on the block of audio-to-MIDI translation…

Now I get it: why all the negativity, supposedly about a DAW.  I just completely disregarded your previous post.  Man, this was quite totally unintentional! 

I just edited my own; no way now does it contradict or supplant yours.

No need now to further bash DAW usage, let’s just see if people like you and I positively can get along somehow  —with even different views, of course about a DAW.  (Smile.)

Amen Ptah Ra wrote:

with an introductory sale price at just $79

Mmmm, with so many already available that are free? Okay, DAW integration for those who want it, but why? I just find DAW a nuisance (that is one of the reasons I like Pianoteq, since it does not require a DAW or some other krappy software like kontakt), but even if one uses a DAW and enjoys it, a Audio-to-MIDI project is a one-off thing, I think, and you don't need a DAW for that. Well, I guess you could use Audio-to-MIDI to get MIDI of a singer or of an acoustic instrument and play that live with accompaniment generated from the MIDI. Maybe. One. Day. I bet this is too slow to be done real time with the necessary sub-10ms latency, and will be for the foreseeable future

There is a learning curve for any DAW, that's for sure, but they are definitely advantages : ultra low latency with Reaper/Pianoteq at 48Khz sample rate / 32 buffer size as Pianoteq Standalone limits it at 64 ,  VST Host as Pianoteq is not a VST host so you cannot load any VST plugin should you need to , Mixing capabilities if you record a session different instrumentalists and use an Audio Interface with various analog inputs , Powerful post editing capabilities within the DAW when any pianoteq session has been recorded.  The Standalone interface  is brillant and 'plug and play' but has limited capabilities vs a  DAW/VST3 setup.

with an introductory sale price at just $79

Mmmm, with so many already available that are free? Okay, DAW integration for those who want it, but why? I just find DAW a nuisance (that is one of the reasons I like Pianoteq, since it does not require a DAW or some other krappy software like kontakt), but even if one uses a DAW and enjoys it, a Audio-to-MIDI project is a one-off thing, I think, and you don't need a DAW for that. Well, I guess you could use Audio-to-MIDI to get MIDI of a singer or of an acoustic instrument and play that live with accompaniment generated from the MIDI. Maybe. One. Day. I bet this is too slow to be done real time with the necessary sub-10ms latency, and will be for the foreseeable future

Now what seems to stand apart from the rest that’s been introduced at this thread is the DAW integration you can get, as Prism is simply used as a plug-in or through ARA inside any DAW that specifically supports it.

Check it out for yourself!

https://youtu.be/lh1r21EtpOQ?si=W5PAJe4AXjj7xUuV

Edited this post which began with the previous “Now you have a new kid on the block!” that didn’t really take into any consideration what came from dv already.

Actually I am not a programmer either, my Linux knowledge is very limited. I did installation following the problems when they were appearing.

The problem every system has some missing packs and others different packs installed.

First I installed the pip (python) software. When I try to install "pip3 install transkun" I got the first problem. It was missing some packs that I don't remember now, but I just ask the A.I. (I just wrote in my mozilla that it has an A.I integration, like chrome has) ,(my system is Debian 13) and follow the instruction.

I follow all the steps of the missing packs and after almost 10 gig of download (I needed to download the A.I. components) I just need to activate the VENV python environments (that was for me the principal point).
The program ran smoothly, but I couldn't run with my gpu mode, only with my cpu mode (more slowly).

I hope these steps give you some ideia.

Thank you. Yes this might help my son to overcome the problems (Same approach by the way - missing packets, missing software, AI helped further. But not up to a successfull end. Yet!) With your input that might change. Best Regards

Beco wrote:

Yujia wrote:

Just to let you know the open-source tool I created, it's actually "the latest AI model" published in 2024: https://github.com/Yujia-Yan/Transkun .

Just a wonderful working Yujia.

I put a Rachmaninoff prelude op 23 No2 in your great piece of software, and it did a terrific job.

Congratulations, it is just gorgeous.

And of course, Pianoteq did a tremendous job for reconstructing the midi file as perfect as the audio file...it is so impressive.

Congratulations too, Pianoteq team.

Best wishes.

How did you manage to install this tool?
I am not realy good with programming stuff, but I asked a well versed programmer (doing things like that for a living) to inmstall this from github. He did not succeed.
Any help from your side?

Thanks in advance

Actually I am not a programmer either, my Linux knowledge is very limited. I did installation following the problems when they were appearing.

The problem every system has some missing packs and others different packs installed.

First I installed the pip (python) software. When I try to install "pip3 install transkun" I got the first problem. It was missing some packs that I don't remember now, but I just ask the A.I. (I just wrote in my mozilla that it has an A.I integration, like chrome has) ,(my system is Debian 13) and follow the instruction.

I follow all the steps of the missing packs and after almost 10 gig of download (I needed to download the A.I. components) I just need to activate the VENV python environments (that was for me the principal point).
The program ran smoothly, but I couldn't run with my gpu mode, only with my cpu mode (more slowly).

I hope these steps give you some ideia.

https://huggingface.co/spaces/Fancellu/...on_to_midi

Downloaded the mp3, turned to midi

https://imslp.org/wiki/10_Preludes,_Op....f,_Sergei)

Link for a week (https://limewire.com/d/tG1VM#WjNbsGESoa)

Yujia wrote:

Just to let you know the open-source tool I created, it's actually "the latest AI model" published in 2024: https://github.com/Yujia-Yan/Transkun .

Just a wonderful working Yujia.

I put a Rachmaninoff prelude op 23 No2 in your great piece of software, and it did a terrific job.

Congratulations, it is just gorgeous.

And of course, Pianoteq did a tremendous job for reconstructing the midi file as perfect as the audio file...it is so impressive.

Congratulations too, Pianoteq team.

Best wishes.

How did you manage to install this tool?
I am not realy good with programming stuff, but I asked a well versed programmer (doing things like that for a living) to inmstall this from github. He did not succeed.
Any help from your side?

Thanks in advance

https://huggingface.co/spaces/Fancellu/...on_to_midi

You can even try the sample mp3s at the bottom

It uses an AI model https://huggingface.co/Genius-Society/piano_trans

https://huggingface.co/spaces/Fancellu/.../README.md

MbcCovers2 nice idea, I will post here the audio and the midi files of Rachmaninoff prelude op 23 no2.

You are right, it's impressive how this technology know when pedals (sustains, una corda) are down or up.

Check this out how great the software of Yujia did.

https://www.mediafire.com/file/14ff14hv...2.mid/file

https://www.mediafire.com/file/0vlncuhz...2.mp3/file

That's pretty

You are right, it's impressive how this technology know when pedals (sustains, una corda) are down or up.

Check this out how great the software of Yujia did.

https://www.mediafire.com/file/14ff14hv...2.mid/file

https://www.mediafire.com/file/0vlncuhz...2.mp3/file

Original: https://www.youtube.com/watch?v=CvUTTAw8Dew

Recording of the midi file played with a Pianoteq K2 preset: https://www.mediafire.com/file/h057exaq....flac/file

The midi itself: https://www.mediafire.com/file/7m7acl1d...y.mid/file

I'll check the rest of the tools people were sharing here later and put some example audios for these too.

Just to let you know the open-source tool I created, it's actually "the latest AI model" published in 2024: https://github.com/Yujia-Yan/Transkun .

Just a wonderful working Yujia.

I put a Rachmaninoff prelude op 23 No2 in your great piece of software, and it did a terrific job.

Congratulations, it is just gorgeous.

And of course, Pianoteq did a tremendous job for reconstructing the midi file as perfect as the audio file...it is so impressive.

Congratulations too, Pianoteq team.

Best wishes.