Skip to content
/ awaaz Public

Proof of Concept for tracking and detecting speech impediments. (Work in Progress)

Notifications You must be signed in to change notification settings

OrthoDex/awaaz

Repository files navigation

Awaaz

Awaaz (आवाज़) (Hindi for 'Voice'), is a Proof of Concept (POC) made to detect and track speech impediments such as stutter and aphasia.

This project is a Work in Progress as I take it through the NSF I-Corps Bootcamp and future programs. Check back or get in touch if you are interested in collaborating! The backend ML pipeline is at https://github.com/OrthoDex/awaaz-backend

Inspiration

The work in this project was inspired by Google's Parrotron. Most of the audio samples used to train the ML model are used from the blog and GitHub repository for Parrotron.

Approximately 7.5 million people in the United States have trouble using their voices. Source Speech Therapy is often hard to find, expensive and subjective. Millions suffer over many years by speech impediments either caused by stroke or acquired via birth such as stutter.

"Spasmodic dysphonia (a voice disorder caused by involuntary movements of one or more muscles of the larynx or voice box) can affect anyone. The first signs of this disorder are found most often in individuals between 30 and 50 years of age. More women appear to be affected by spasmodic dysphonia than men." Source

For patients not covered by health insurance, speech therapy typically costs $200-$250 for an initial assessment, then about $100 to almost $250 per hour. Source Speech therapy can easily be over 2 years causing significant levels of money being spent.

Bringing affordable speech impediment diagnosis and therapy will democratize this common problem and give access to low cost therapy to everyone. It'll also allow users to track their speech "scores" over time to catch any relapses and help doctors to check improvements.

What it does

A simple frontend app records your voice and posts it to a backend app. The app runs multiple signal processing tasks such as spectral rolloff, Mel-frequency cepstral coefficients and centroid computation. These scores are then fed into a machine learning algorithm trained on a dataset of speech impediment audio recordings and their synthesized versions. The backend app also uses Google's Speech to Text to transcribe the audio and Text To Speech to get a machine benchmark for voice. A person's voice metrics are gauged against the synthetic voice that's generated by the combination of Google Speech to Text and Google Text To Speech.

These benchmarks form the basis of an ML classifying model that returns back the certainty of a speech impairment. These metrics are also surfaced to the user to allow them to track their speech therapy over time.

Tech used

  1. Google Cloud AutoML
  2. GCP Speech to Text and Text to Speech apis
  3. Hasura GraphQL
  4. React
  5. Pandas and Principal Component Analysis analysis using scikit-learn.
  6. Auth0 for Google SSO

What's next for AwaazApp

  1. Better classification scores
  2. Better UX to track speech therapy

Demo and other sources

The Devpost link for this project includes the temporary demo link hosted on Glitch.

About

Proof of Concept for tracking and detecting speech impediments. (Work in Progress)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published