Ideally, it would be voice.mozilla.org. But we need to figure out how to get this.

Figure out scope of testing, writing into SOW, get over to Softvision for quote

We'll need to do client detection and server conversion.

Perhaps one thing we could do is add extension to the url: /upload/random.m4a to give the server a hint about what to serve.

Better though would be client side audio conversion. Need to see what is easier, converting ogg -> m4a on webkit, or the other way around on everything else, or neither.

Is it possible for us to publish a "wrapper" app on the app store, and then change the content later? If so, let's build it!

Right now the app displays underneath the system statusbar (which is transparent). This looks a little weird. Can we make it so that the app displays inside a frame and not underneath the statusbar?

Landing Page (main)

• Show overall progress towards goal
• Big Donate now button

Leaderboard

• show top 3
• show where you are
• show bottom 2

Rewards Page

• List any rewards the user has (with unclaimed ones at the top)
• Ability to add name for leaderboard, new sentence, login

Contributors

• Dynamic list of all users who reached level 2

Voice Donation Page

• experience points (voice-coins), perhaps count of how many submissions from your user
• Next reward widget - a

Profile Page

• Whose budget?
• What hosting company?
• Security review?

It is important that the submission functionality has a pleasant effect, since we want people to do this often.

right now we just show a blank screen. i have a feeling this will be a show stopper for the apple review

We need a final decision and set of issues/user stories generated on what specific stickiness/gamification functionality we are building for the June Activation release.

@kdavis-mozilla is the final decision-maker

The rest of us will build out options that he can sign-off.

Quickly mock-up (not code) and get feedback on stickiness/gamification options before we make a final choice.

Depends on building out a long-list of options.

There are several issues here I'm not sure if I should break them up in different issues.

• The general flow should be the user only needs to press on Yes / No to move on to the next sentence. No separate Submit button.
• Instead of the browser sound controls, we should have a Play button, which becomes Repeat after the first time it is pressed.
• After submitting, the next sentence should be loaded and the audio played automatically. The goal is to have as few user interactions as possible.
• A user should not hear the same (sentence, audio) pair more than once.
• There seems to several types of recorded audio samples:
  • Accurate readout of the sentence
  • Medium quality, perhaps a word or two are missing
  • Nothing was recorded at all
  • Background noise
  • Something else was recorded entirely, perhaps the used accidentally pressed the record button and left it on, posing a privacy concern.
  • Profanity, or vulgar content that should be flagged.

Dependant on draft of architecture in issue #13

Getting an error from the audio element.

Right now, we only have 3K sentences or so.

Two possible new sources are

1. Wikimedia text
2. BYU Corpus
3. Leipzig Corpora

Home page:
Build the world's most diverse set of voice data that researchers and others can use for free to create better voice technologies for the Internet.

[maybe button here?]
Click 'Record' to start donating your voice.

Your voice donations will be made available for researchers and others to use under a Creative Commons license [https://creativecommons.org/publicdomain/zero/1.0/]. Your name or any other identifying information will not be associated with this voice data.

This project is governed by Mozilla's Privacy Policy [https://mozilla.org/privacy/websites/]

About page:
Project Common Voice is brought to you by Mozilla, the proudly non-profit champions of the Internet.

Today's technologies that allow learning from data are freely available for anyone to use, and are resulting a wave of innovation online. However, voice technologies (for example, speech recognition) are not seeing the same innovation because little data is freely available to train machine learning technologies. The data that is available is from a set of speakers with limited diversity of accents and languages.

Our aim with Project Common Voice is to enable "voice donors" to build the world's largest and most diverse set of voice data that is freely available for anyone to use. Our vision is that researchers and others will be able to use this data to increase innovation in voice related technologies. This will help everyone have access to a new wave of voice technologies, and ensure that people aren't locked-in to using services from a small number of Internet giants.

If one goes to the website without ever previously visiting the site, one hits the error

Uncaught TypeError: Cannot read property 'userId' of null
    at User.Component.setState (bundle.js:404)
    at User.restore (bundle.js:487)
    at new User (bundle.js:475)
    at new App (bundle.js:1430)
    at HTMLDocument.<anonymous> (bundle.js:301)

due to restore() incorrectly initializing userId on first time use.

Users can help us validate voice input. Right now we don't have this scoped, so we would need to do that if this becomes a priority.

This was originally brought up in issue #3.

• flow-chart to describe data structures and relationships

• legal review per issue #12

Need a UX/UI contractor for roughly 1-2 weeks for interaction and visual design of the responsive website layout.

• Bounce rate at various entry points
• First time experience funnel

We know we at least need the following fields:

• email address
• userid (generated in the client, separate from collection id)
• accent
• year of birth
• gender
• display name

Decision based on advice from legal on the license type for the data sets that will be made public.

Reach out Selena D

• iOS build to bsmith for legal review and identifying issues - @andrenatal

• feedback from bsmith --> leads to issues to incorporate

Right now you can drag the content of the webview up and down due to Safari's overscroll ability. Can we disable this in the app?

which is bad because we use userid in the audio file path

Need a production environment on IT managed AWS account. Ensure that is setup so that we can prod there.

Here's a proposal for a simple first time experience that introduces all the features. Obviously this is still open to suggestions and/or complete rewrites.

User starts at Homepage:
(Text needs to be short and clear, and have call to action.)

Welcome to Voice Commons!
Our goal is to create a Public Domain (CC0) database of voices in every language with any accent. We believe this data can empower universities, researchers, buinesses, non-profits (like ourselves), or anyone who is interested in Voice Recognition technology to be creators in this emerging space. You can learn more about Common Voice through our Mision Statement.

Take our guided tour:
Button text could be Lend your Voice or Donate or Try it out

(button takes you to donate screen)

Donate Screen

insert brief description of why we need "labeled" voice clips.

(pehaps we can call out that your data won't be uploaded yet)

User clicks records their voice, and clicks Submit.

(here we could have some sort of nice indication they just contributed)
(submitting their voice clip takes them to the listening screen.)

Listen Screen

Explain that we need to make sure that what everyone says is correct
"Play this sound clip, and tell is if what they said matches the senstence"

(here users could listen to your own clip, or we could have a pre-canned example)

User selects yes or no:
Yes || no

(Whether they say yes or no, move on to registration)

Don't worry, what you just recorded has not been added to our database yet. First, we would like you to take a look at our Privacy Policy.
tl;dr When we publish the voice database, we will strip all personally identifying information about you. But while using this site, you may share as much or as little personal information as you'd like with other users.

To use all the features of this site, you need to register. Again, your email will not be tied to any published voice data.
Registration Form
(perhaps make it clear they can still donate as Guest)

This is ok right now since we restart the server frequently. But this will need to be fixed before release.

We could even attach the permission prompt to the first time I call "start_capture".

Brian will double-check Wikipedia and Gutenberg

When clicking on the "Next" button on iOS it remains grey after the next sentence has appeared