About Mozilla Common Voice

Voice is an increasingly dominant way to interact with technology. Currently, lots of people are being left behind. Mozilla Common Voice wants to change that by mobilising people everywhere to share their voice.

Mozilla Common Voice is Mozilla’s flagship initiative to de-bias AI by mobilising real people everywhere to share their voice. With almost half a million contributors, 76 languages and over 15,000+ hours of voice data, Common Voice is the largest publicly available voice dataset in the world.

What we’re looking for

The Common Voice platform allows people to record themselves reading short sentences. Others can then play your sentences to check they can be heard and understood, and these clips are made available for anyone to use through a regular dataset release. The major contribution components of the platform are a Speak interface and a Listen interface.

The platform is a modern React web application. For it to function effectively it requires a stable, strong internet connection. You can read more about the technical specifications of the MCV platform here.

We know that many people do not have access to powerful, reliable internet connections - and we want Common Voice to be open to everyone.

We are excited to announce that we are opening a call for proposals for Low & No Connectivity Contribution Solutions. This call is made possible by our partnership with NVIDIA.

Category 1.

Make Common Voice into a progressive web app

Size: Up to $5000

Category 2.

Safe, secure contribution via a mobile application (new app, or leveraging existing or third party apps)

Size: Up to $5000

Category 3.

Open call - proposals that would enable offline or low bandwidth contribution that do not fit into either of the above categories.

Size: Up to $5000

Criteria

Your solution must:

  1. Make it easy for people in low connectivity environments - for example those reliant on low quantities of mobile data, or those with unreliable connections, or slow internet speed - to contribute their voices (Speak)
  2. Make use of the existing Common Voice platform and infrastructure and contribute directly to the Common Voice dataset
  3. Respect the privacy and security of all contributors, for example by only leveraging messaging platforms with end-to-end encryption by default
  4. Be thoughtful about the user experience of a diverse range of contributors, including those with different levels of technical and digital literacy, different levels of context about the Common Voice project etc
  5. Make significant efforts to comply with web accessibility guidelines
  6. Be made open source

Your solution could:

  1. Make it possible for people with no internet connection to contribute (offline contribution)
  2. Make it easy for people with no/low bandwidth to validate voice clips clips (Listen)
  3. Fully meet web content accessibility best practice (eg, be WCAG 2.1 AAA compliant)

Evaluation process

Your proposals will be judged by the following panel:

Jenny Zhang - Senior Engineer for Common Voice

Em Lewis-Jong - Product Lead for Common Voice

Chenai Chair - Senior Advisor for the Africa Mradi

Francis Tyers - Linguistics Advisor for Common Voice

Ben Hohner - Engineering Manager for Mozilla Foundation

Who can apply

  • Individuals or small communities and teams
  • Small or medium sized enterprises (fewer than 50 employees)
  • Non profit organisations

We welcome existing Common Voice community member applications.

How to apply

Submit your proposal to the panel via the Product Lead at [email protected]

It should give details on:

  1. Product and design intentions
  2. Intended technical implementation
  3. Thoughts on ongoing maintenance and sustainability
  4. Timeline
  5. Team and skills
  6. Metrics for success
  7. Financial amount requested
  8. Any other details on how your proposal meets the criteria

Deadline: January 5th 2022