Onsets and Frames Transcription - JavaScript vs. Colab

[Fokko Breden]

I am currently converting some audio files with onset on frames. I noticed that I got different MIDI results with the JavaScript app than with the Colab notebook. Do they use different checkpoints or is it due to some parameters?

[Curtis "Fjord" Hawthorne]

They actually use different models. The javascript one is smaller and has a forward-only LSTM for efficiency. You'll likely get more accurate results by using the colab or command line version.

-Fjord

On Tue, Mar 9, 2021 at 8:29 AM Fokko Breden <fokko....@gmail.com> wrote:

I am currently converting some audio files with onset on frames. I noticed that I got different MIDI results with the JavaScript app than with the Colab notebook. Do they use different checkpoints or is it due to some parameters?

--
Magenta project: magenta.tensorflow.org
To post to this group, send email to magenta...@tensorflow.org
To unsubscribe from this group, send email to magenta-discu...@tensorflow.org
---
To unsubscribe from this group and stop receiving emails from it, send an email to magenta-discu...@tensorflow.org.

[Fokko Breden]

... Thank you for the information. And yes, I get more detailed MIDI files generated with Colab.

[Prashanthi Atukuri]

Hi Fokko,

can we connect sometime to discuss this in detail. We are presently working on a similar project and could use some deliberation from you.

Thanks

Prashanthi

[Fokko Breden]

Hi Prashanthi,

Unfortunately, I am not familiar with AI things. I use Magenta's models pragmatically - as they are.

Fokko