Building and Implementing a Deep Learning Speech Classifier in Python

Curious to learn about working with audio data or machine learning? In this workshop we learn about how to prepare audio data for machine learning and apply learning algorithms to it.

We will lightly cover the background of sound/speech and neural networks such as convolutional neural networks and long short-term memory networks. Our main goal, however, is to get a hold of speech data, prepare it for training, train models, and finally to test the models on new speech.

While some of the code is a bit complex, this workshop is meant for everyone. If you are new to Python, we will walk through the code together and there will be people there to help! You will walk away with new Python tricks and a perspective into fun projects using Python so please don't shy away.

• Audience level:
Intermediate (Familiarity with basic Python and Machine learning concepts to get most out of it, Beginners mostly welcome, preparation material will be announced soon)

• Speaker
Aislyn Rose
https://github.com/a-n-rose

Bio
Aislyn's official education is based in research. She earned a bachelor's degree in psychology and a master's degree in clinical linguistics, which is basically the study of how language works in the brain. She focused on human learning of new languages, specifically how the brain handles more and less complex acoustic information. For example, does hearing new words from a variety speakers (not spoken the same time!) help new language learning? Or is it better to hear new words from just one speaker?

When Aislyn was 25, one professor introduced her to programming, which she hated at first. Eventually she started having fun with it and actually shifted career directions. She now pursues a career implementing both development and research. To her delight, a hot-topic in current research is the application of deep learning models to sound and speech. Note: deep learning was not a topic covered in her academic education. That she gained through self-study.

A child and a few years of self-study later, she dedicates her skills to opensource projects and teaching Python. She is currently a developer for the opensource project NoIze. NoIze aims to improve hearing aids and noise processing with the implementation of artificial intelligence.

• Host
omni:us - Supercharging insurance with AI
https://omnius.com/

• Requirements
Please follow the instructions and have Python 3, virtualenv, required packages needed and DOWNLOAD the dataset beforehand!
https://github.com/a-n-rose/Build-CNN-or-LSTM-or-CNNLSTM-with-speech-features/blob/master/INSTALLATION.md

For windows or any issues with installation please reach out in our slack in channel #troubleshooting (check below)

• Miscellaneous
Drinks and pizzas will be available.

• Gender policy
We believe knowledge is for all and at the same time our events aim primarily to empower women tech community. We request non female attendees to be aware of these situation and make their presence discrete. Eg. by coming with a female plus one to ensure gender balance, avoiding to be heard more than the rest of the attendees in discussions and question sections.

• Photography / video consent
We take photos and videos during the event to use for documentation and in social media such as here in Photo albums, Facebook, Twitter, etc. By coming to the meetup, you willingly give consent to take photos and videos of you. If you do not want to give your consent, please let us know at check-in.

• Code of Conduct
By attending one of our events you agree to follow our code of conduct. You can read it online here - https://www.pyladies.com/CodeOfConduct/

• Contact
Interested in speaking at one of our events? Have a good idea for a Meetup? Get in touch with us at berlinpyladies@gmail.com

You can also find us on slack
Invite: https://pyladies-berlin.herokuapp.com/
Slack: https://pyladies-berlin.slack.com

Disclaimer:
Photo by mozilla website - https://research.mozilla.org/machine-learning/