james-trayford / strauss

Sonification Tools and Resources for Analysis Using Sound Synthesis
Apache License 2.0
33 stars 4 forks source link

Tts #11

Open SamYoules opened 1 year ago

SamYoules commented 1 year ago

Basic caption functionality, with caption placed at the beginning of a sonification. Jupyter notebook is in tests/pre-merge.

james-trayford commented 1 year ago

Great, thanks @SamYoules !

Think we will want to merge PR 10 first, will take a look in the meantime. We can rebase this one afterwards

I noticed there were some python compile files (.pyc) which the gitignore changes from the other branch should filter out. A couple of checkpoints got checked in e.g. (examples/.ipynb_checkpoints/AudioCaptionTutorial-checkpoint.ipynb) which is something it looks like would be good to add to the .gitignore for the future too. Also the notebooks still seem to have output, so the pre-commit hook you're using doesn't seem to be doing it's job properly. Will look into that too.

excited to try it out!

SamYoules commented 1 year ago

The python compile files and notebook checkpoints were probably from the previous commit.

james-trayford commented 1 year ago

yeah, so we can just git rm the files we don't need in a new commit to this branch, and then future hooks and .gitignore files should filter these out in the future

james-trayford commented 1 year ago

@SamYoules a few edits to:

We'll want to rebase this onto the main once PR #10 merged, and clear up unneeded files.

Can we adapt the tutorials/ notebook into the examples/ directory with the others? could use something similar to the test/pre-merge/ notebook where all the TTS is done in the context ofstrauss captions

james-trayford commented 1 year ago

was also wondering whether we should merge this into the main once this functionality is added , or also implement the basic audio 'key' aspect too... lmk what you think!

james-trayford commented 1 year ago

I rebased the branch to incorporate repo-management changes (PR #10)

SamYoules commented 1 year ago

soni.render sometimes crashes at line 191 self.caption_channels[str(i)].values += (wavobj*cnorm*panenv) with the error: ValueError: operands could not be broadcast together with shapes (227599,) (227600,) (227599,) Possibly a rounding error on wavobj?