robz
commented
2 years ago Here is a similar topic I'd be happy to present on. It's a bit more basic than what you described above, but could be a good stepping stone towards more advanced topics.
Musical representations
Context
Just as we use text to encode spoken words, we also have several ways to encode music, such as sheet music and the MIDI file format. These musical representations can be adapted to use for machine learning applications that involve processing and generating music.
Duration
10 minutes
Talk flow
- Motivation - why musical representations are useful
- Sheet music, scores, piano rolls
- MIDI, performance data
- Learned representations
Key takeaways
The audience will get an understanding of several musical representations, how to use them to build models for music-related tasks, and pointers to helpful frameworks and papers to learn more.
Vaibhavs10
JonathanSum
Context
AI-generated music is not new, however, the recent advancements in TTS have led to more fine-grained control over the generated sound.
Duration
10 minutes
Talk flow
Key takeaways
The audience will get an understanding of TTS and how can one tweak the architecture to gain prosodic control over the generated speech.
P.S. The other details above are just a suggestion and you can tweak it at your convenience, just copy this template and respond with what you would like to cover.
If you would like to present on this topic or suggest a speaker, please leave a comment below :)