Machine Learning Techniques for Audio and Speech synthesis

[kdhingra307]

• Abstract This talk is about how audio and speech synthesis differs, how it has evolved from the last couple of years with the deep learning techniques. I will be going through both statistical and neural techniques for audio synthesis with a comparison based on performance and quality

• Brief Description and Contents to be covered Audio and Speech synthesis

• Pre-requisites for the talk

• Time required for the talk

• Link to slides

• Will you be doing a hands-on demo as well? yes, I can

• Link to ipython notebook (if any) Will upload it by the end of the day

• About yourself I am a final year undergraduate at Cluster Innovation Centre. Currently, I am working on fast speech pipelines at IIIT-Delhi as a research project and also contributing at WikiMedia as Google Summer of Code

• Are you comfortable if the talk is recorded and uploaded to PyData Delhi's YouTube channel ? yes

[manojpandey]

@Dawny33 Please review sir ;)

[Dawny33]

The topic looks pretty interesting.

@kdhingra307 Pl add more details regarding how you want to structure the session.

[kdhingra307]

Hi @Dawny33 yaa sure

this is kind of overview of what would be comprehensible for most

1. A brief overview of human-computer interaction, why is it so important.
2. Quick basics into signals and what are the major differences between audio and speech synthesis. Intro to statistical and neural vocoders.
3. Is neural powerful or statistical?? comparison between two (will be using wavenet and griffin-lim)
4. Ongoing work in the field of Speech processing (will include tacotron-2 demo and quickly go through the tacotron-2 model)
5. What I am currently doing(making tacotron-2 faster and near real-time)

your suggestions are most welcome

[Ridhwanluthra]

@kdhingra307 any update on the slides and jupyter notebook? We are planning a meetup on 16th and would love you have your talk then.

[kdhingra307]

@Ridhwanluthra I will add those slides in 1 or 2 days

I would love to give the talk at the next meetup

[kdhingra307]

Hey guys

I have attached a link to the GitHub repository which contains summary.pdf, this mainly includes the subjects I will be specifically sharing tomorrow

But in general, I will use jupyter notebook, as these will allow an interactive communication. I have yet to add the notebook in the repository, will add them in a couple of hours and update you guys too.

https://github.com/evolution_of_audio_and_speech_synthesis

[manojpandey]

@MSanKeys963 can we slate this for next meetup?

[MSanKeys963]

This was delivered in the last meetup. I've added the appropriate labels.