Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Mozilla Overhauls Speech-To-Text Contribution Interface

voice.mozilla.org

Post reply on HN
Print view View original

31–40 of 46 posts

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#31

Post by a_imho »

userbinator wrote: post #19

From the description: Common Voice is a project to help make voice recognition open to everyone. Now you can donate your voice to help us build an open-source voice database that anyone can use to make innovative apps for devices and the web. I'll be the first to note that here's another piece of personally identifying information you just "donated"...

I won't contribute because speech interfaces are imo terrible and against the proliferation of Echo and Duplex like services, but the data collected is listed here

https://voice.mozilla.org/en/privacy

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#32

Post by adrianN »

a_imho wrote: post #31
userbinator wrote: post #19

From the description: Common Voice is a project to help make voice recognition open to everyone. Now you can donate your voice to help us build an open-source voice database that anyone can use to make innovative apps for devices and the web. I'll be the first to note that here's another piece of personally identifying information you just "donated"...

I won't contribute because speech interfaces are imo terrible and against the proliferation of Echo and Duplex like services, but the data collected is listed here https://voice.mozilla.org/en/privacy

Speech interfaces have some limited usecases where they are much better than alternatives. For example in situations where you need your hands for something else, like in a car.

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#33

Post by punchingwater »

ipsum2 wrote: post #7

This is a nicely designed interface. Well done, Mozilla. Validating sentences is quite fun, listening to different accents from around the world. Try it out if you haven't already: https://voice.mozilla.org/en/listen It's awesome that the dataset is offered with a CC-0 license: https://voice.mozilla.org/en/data , does anyone know if it includes the answers from the survey? I have a limited bandwidth internet, so I ha…

We used some of the research around Mechanical Turk to find best practices for limiting trolling (e.g. [1]). Our approach thus far has been the two-thirds rule: if two out of three people say the clip is good/bad, we trust that. Also note, that we have seen remarkable low trolling numbers, and that most of the invalid are pronunciation mistakes or saying the wrong word.

1.) https://groups.csail.mit.edu/sls/publications/2010/McGraw_LR...

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#34

Post by punchingwater »

archgoon wrote: post #28
archgoon wrote: post #24

Earlier quoted context omitted.

> It's awesome that the dataset is offered with a CC-0 license: https://voice.mozilla.org/en/data , does anyone know if it includes the answers from the survey? I'm downloading it now, I'll have an answer in a half hour. Does anyone know if there is a torrent for it?

Well, download took longer than expected :). Anyhow, here's a sample from the csv file: filename,text,up_votes,down_votes,age,gender,accent,duration cv-valid-test/sample-001224.mp3,but i felt miserable watching him wither away like a shriveled dandelion,1,0,thirties,male,england, Not sure how some of these are being populated, but yeah; there's several additional folders including invalid mp3, a splintered train set…

We also keep the README in the repo: https://github.com/mozilla/voice-web/blob/master/docs/corpus...

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#35

Post by punchingwater »

makmanalp wrote: post #25

This is wonderful and addictive. One thing that comes to mind is that the UI allows for very little metadata - for example in some cases the audio has a slight mispronunciation even though the intended word was clear - wouldn't it be helpful to mark "difficult" cases like this? In other cases the volume is just super low or there is background noise. The other thing is that it's very cool to see the "you helped us re…

We do have a issue filed to allow users to tag recordings with certain metadata, like noisy or male/female voice. https://github.com/mozilla/voice-web/issues/814

It is something we are still working on.

> The other thing is that it's very cool to see the "you helped us reach out x% goal" thing but it locks up all the previous / next shortcuts which means I have to switch back to the mouse after 5 entries.

That's a bug! Would you mind filing one here: https://github.com/mozilla/voice-web/issues

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#36

Post by punchingwater »

mgkimsal wrote: post #29
makmanalp wrote: post #25

This is wonderful and addictive. One thing that comes to mind is that the UI allows for very little metadata - for example in some cases the audio has a slight mispronunciation even though the intended word was clear - wouldn't it be helpful to mark "difficult" cases like this? In other cases the volume is just super low or there is background noise. The other thing is that it's very cool to see the "you helped us re…

> for example in some cases the audio has a slight mispronunciation even though the intended word was clear had similar issue/concern. ideally if enough people mark something as correct, the variations and slight differences will get merged together. it did still bother me a bit, as being able to add a bit more extra data would probably be helpful. but... maybe they can add some geo-ip data - respondents from various…

> Also... (too lazy to check right now) - if I create an account, can I see the 'yes/no' ratings of my own submissions?

Not yet, but this is something in the works. You can explore our new experience with the evergreen link: http://bit.ly/cv-desktop-ux

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#37

Post by punchingwater »

redfast00 wrote: post #18

Am using Brave on Android (basically chrome with afblocker). I accidentally mislabeled some (probably) correct samples because the waves move when you click the play button, even if the audio hasn't loaded yet and I thought it was just a blank recording.

Would you mind filing an issue? https://github.com/mozilla/voice-web/issues

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#38

Post by punchingwater »

polygot wrote: post #26

Really nice app. It would be a nice feature to add volume normalization as some microphones/speakers are very soft, and I can't hear what they are saying, while others are much too loud.

There has been some discussion around this, but no real movement yet: https://github.com/mozilla/voice-web/issues/336

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#39

Post by punchingwater »

amelius wrote: post #6

Did they consider to use MTurk for this?

In the early days of this project, before we shipped the website (ie. ~March of 2017), we did some explorations around Mechanical Turk. The problem with the Mech Turk approach is that for recording your voices you need a lot of different people speaking (ie. 10s of thousands). But for languages other than English, Mech Turk simply doesn't have these kind of numbers. And indeed English is not that interesting to us, since there exists public data already in English (see LibriSpeech). There are of course other micro-task platforms popular in other countries (for instance, there's a myriad in Indonesia), but we didn't have the time to manage jobs on all these different platforms.

However, Mech Turk is better for things like validation, since you only need a handful of people doing the majority of work.

In any case, I have some very hacky tools we used for this exploration, if you are interested: https://github.com/mikehenrty/mech-turk/

Re: Mozilla Overhauls Speech-To-Text Contribution Interface

#40

Post by punchingwater »

userbinator wrote: post #19

From the description: Common Voice is a project to help make voice recognition open to everyone. Now you can donate your voice to help us build an open-source voice database that anyone can use to make innovative apps for devices and the web. I'll be the first to note that here's another piece of personally identifying information you just "donated"...

Thank you for bringing this up.

Indeed Common Voice is not for everyone. We try to make it clear in our Privacy Policy [1] what pieces of data we collect and why. We do not publish email address or names with the data, and we even strip speaker identification info (so that a speaker's recordings are not grouped but instead everyone's recordings go into one giant bucket). That said, if this still makes you feel uncomfortable, we understand. And if you would like to contribute without donating your voice, you can always validate the recordings of others.

1.) https://voice.mozilla.org/en/privacy

Post reply on HN