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

Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

blog.mozilla.org

Post reply on HN
Print view View original

41–50 of 90 posts

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#41

Post by skykooler »

punchingwater wrote: post #33
hardwaresofton wrote: post #30

For those who don't know that open source speech recognition that does not depend on AI already exists: http://cmusphinx.sourceforge.net/ http://julius.osdn.jp/en_index.php Maybe with this data set released eventually all that additional data will be used to improve those tools as well

Don't forget Kaldi! https://github.com/kaldi-asr/kaldi

The problem with Kaldi is that it's virtually impossible to get a dictation model working with Kaldi unless you have a doctorate in speech recognition. There is no "I know basic programming, but little about speech recognition" documentation for Kaldi.

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#42

Post by woodson »

skoocda wrote: post #5
tmzt wrote: post #2

How does this compare with Snips.co which can do offline speech recognition on a Rapberry Pi 3? Coyld this be used to train a model/engine that can be used that way?

I can't find exact numbers on Snips.ai, but generally there's a linear relation between the size of the inference model in RAM and the accuracy it can obtain. I'd have to assume DeepSpeech outperforms anything running on a RasPi3, at least for LVCSR. It hits 93.5% accuracy on Librispeech, which I've never seen from any offline recognition models.

Kaldi has 4.14% WER (95.86% accuracy) on the same test dataset (test-clean) [1] using a model that runs faster than real time on CPU. You would have to make the model smaller to run it in real time on a RasPi3, but according to this [2], you can get decent WERs for read speech even then.

[1] https://github.com/kaldi-asr/kaldi/blob/master/egs/librispee...

[2] https://groups.google.com/d/msg/kaldi-help/Pr6jPH1Qshg/kn8df...

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#43

Post by tim-- »

punchingwater wrote: post #27
gok wrote: post #22

This is super cool, but I'd be cautious about the usefulness of this data set. Both this data set and LibriSpeech are read speech, where the speaker was prompted with a transcription and asked to say it out loud. In practice it's very rare that you're trying to transcribe speech that's already been transcribed. Speech patterns for computer-directed speech (e.g. for voice activated user interfaces) or human-to-human s…

Yup, this is an excellent point. We have, and will continue to explore ways to allow Common Voice users to speak more organically (for instance by answering a question, or responding free-form to some other sort of prompt). The problem with this approach is that it requires an extra step, transcription, which at the scale we are trying to achieve is pretty costly in either money or time (ie. tedium for our users). Ev…

I was thinking earlier that maybe YouTube CC-licensed audio with manually entered subtitles might be a good source?

Though, most videos of decent length would only contain say three or four speakers, which is most definitely sub-optimal.

https://www.youtube.com/results?sp=EgYYAigBMAE%253D&search_q...

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#44

Post by nullc »

tim-- wrote: post #43
punchingwater wrote: post #27

Earlier quoted context omitted.

Yup, this is an excellent point. We have, and will continue to explore ways to allow Common Voice users to speak more organically (for instance by answering a question, or responding free-form to some other sort of prompt). The problem with this approach is that it requires an extra step, transcription, which at the scale we are trying to achieve is pretty costly in either money or time (ie. tedium for our users). Ev…

I was thinking earlier that maybe YouTube CC-licensed audio with manually entered subtitles might be a good source? Though, most videos of decent length would only contain say three or four speakers, which is most definitely sub-optimal. https://www.youtube.com/results?sp=EgYYAigBMAE%253D&search_q...

The last time I checked Youtube's terms of service prohibit you from making use of the rights granted by the creative commons licenses on the content.

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#45

Post by woodson »

asabjorn wrote: post #12

Does this relate or help at all with speaker identification? The Microsoft speech API also provides speaker recognition which is useful for many applications: https://azure.microsoft.com/en-us/services/cognitive-service...

No, this is for speech recognition only.

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#46

Post by peteretep »

gok wrote: post #34
punchingwater wrote: post #27

Earlier quoted context omitted.

Yup, this is an excellent point. We have, and will continue to explore ways to allow Common Voice users to speak more organically (for instance by answering a question, or responding free-form to some other sort of prompt). The problem with this approach is that it requires an extra step, transcription, which at the scale we are trying to achieve is pretty costly in either money or time (ie. tedium for our users). Ev…

Have you considered getting volunteers to transcribe permissively licensed video or podcasts?

One advantage of being the size / prestige of Mozilla is presumably organisations that are willing to license their content for free to Mozilla for this purpose?

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#47

Post by tonmoy »

muxator wrote: post #9

I am very grateful for this release from Mozilla, and more generally for the broad vision of their effort. As time passes, the quest for openness and freedom in software moves higher up in the stack. Thanks to the latest ~30 years of effort, we basically came to a point in which we have free OSes, basic infrastructure, building tools, end-user applications. In the last ~10 years we changed paradigm: autonomous deskto…

It’s exciting to see open source RISC-V gaining momentum on the HW world

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#48

Post by mkbkn »

zeep wrote: post #26
api wrote: post #18

Earlier quoted context omitted.

Everything by Google sends everything to Google.

but most useful keyboards require permissions that are scary...

Have you checked Multiling O Keyboard app? (I used it more than 1 year ago, now I use a phone powered by Sailfish OS).

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#49

Post by visarga »

punchingwater wrote: post #27
gok wrote: post #22

This is super cool, but I'd be cautious about the usefulness of this data set. Both this data set and LibriSpeech are read speech, where the speaker was prompted with a transcription and asked to say it out loud. In practice it's very rare that you're trying to transcribe speech that's already been transcribed. Speech patterns for computer-directed speech (e.g. for voice activated user interfaces) or human-to-human s…

Yup, this is an excellent point. We have, and will continue to explore ways to allow Common Voice users to speak more organically (for instance by answering a question, or responding free-form to some other sort of prompt). The problem with this approach is that it requires an extra step, transcription, which at the scale we are trying to achieve is pretty costly in either money or time (ie. tedium for our users). Ev…

> could voice somehow be a "progressive enhancement" to the web?

I have created my own TamperMonkey plugin that adds TTS to web pages. It finds text, makes it clickable, and when a user clicks a word, it starts reading from there, highlighting text as it reads it, skipping menus and chrome. I find this helps me better focus on reading. Unfortunately I can only stand one single voice and it's been stagnating for years (Alex from Mac OS). Can't wait to hear the WaveNet voice Google has been threatening to give us.

Re: Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data

#50

Post by yjftsjthsd-h »

RubenSandwich wrote: post #37
punchingwater wrote: post #29

Earlier quoted context omitted.

Thank you so much! I also want to emphasize the importance of listening (validating) as well as recording. Validation is an big part of the puzzle for building machine learning viable data.

Can I suggest is encourging user's to get recordings from their children as well, as most speech recognition libraries are pretty poor with children's voices. (IMO Alexa Voice Service is by far the best with children voices.)

Is that okay legally? Maybe parental permission is enough
Post reply on HN