adrianbg

8 ms·

31.
by adrianbg 9y ago
That's fair. I think both approaches are useful in different situations. Mozilla seems to be focussing on Siri-like use-cases as opposed to dictation. Even for dictation, for many people having to train the system themselves is more wo
32.
by adrianbg 9y ago
Yeah. I don't get the impression that the Kaldi core team has been trying very hard recently to get SOTA on eval2k/switchboard. This number uses one acoustic model with a trigram LM decode + fourgram rescoring -- there isn't
33.
by adrianbg 9y ago
Here are some benchmarks on telephone speech, including both APIs and human transcription services: https://remeeting.com/app/benchmarks Google actually did pretty badly for us on extended telephone speech. Not sure wh
34.
by adrianbg 9y ago
I agree. I think a big part of the reason that people use WER is that it's relatively unambiguous and easy to measure.
35.
by adrianbg 9y ago
Totally. You may want to take a look at the papers from CHiME4 for more along those lines: http://spandh.dcs.shef.ac.uk/chime_challenge/chime2016/resul... I'm really fascinated by the whole idea of blind sour
36.
by adrianbg 9y ago
The 8.5% in this file is what you'd compare to Microsoft and IBM's recent ~5% results. https://github.com/kaldi-asr/kaldi/blob/master/egs/fisher_sw... Kaldi hasn't been in first place
37.
by adrianbg 9y ago
Unfortunately that Tensorflow integration didn't include acoustic modeling, so you still need to use Kaldi's neural net toolkit for that.
38.
by adrianbg 9y ago
Exactly. This is also why Google sponsors the CHiME challenge, the existence of which is more proof that ASR is pretty far from solved. http://spandh.dcs.shef.ac.uk/chime_challenge/
39.
by adrianbg 9y ago
The problem wrt open source / free solutions is data. Kaldi is open source and gets state of the art results -- but the data costs a lot of money. Training the models is doable on a commodity GPU although it takes quite a while.
40.
by adrianbg 9y ago
Hah sure. By "hard," I meant that it's the largest hurdle. And probably even the common research datasets aren't enough to give you results competitive with Google, etc. AFAIK Google uses its own hand-transcribed data.
41.
by adrianbg 9y ago
Although some models are end-to-end neural nets, most of the ones in production (and all of the ones that get state of the art results) only use a neural net for one part of the process. Lots of people are as surprised as you, but that&#x27
42.
by adrianbg 9y ago
You are correct that the problem is data. Kaldi is hard to use, but making it easier to use isn't as hard as getting good training data. Mozilla's project is a good start for some purposes. One flaw with it is that they're ha
43.
by adrianbg 9y ago
Speech to text is exactly speech recognition. Speech to text isn't solved.
44.
by adrianbg 9y ago
The low WER numbers you've probably seen are for conversational telephone speech with constrained subjects. ASR is much harder when the audio source is farther away from the microphone and when the topic isn't constrained.
45.
by adrianbg 9y ago
If you think that's bad, check out the Geordie accent from Newcastle: https://www.youtube.com/watch?v=ZY4TT3VtR8o According to Wikipedia it's "a direct continuation and development of the language spoken by t
46.
by adrianbg 9y ago
CMUSphinx is really old. Kaldi is hard to use, but it's much better.
47.
by adrianbg 9y ago
Alexa's recognition is made more difficult by the fact that you talk to it from a distance. Still, though, you're right that only handling predefined commands should make it much easier.
48.
Show HN: Serverless speech recognition with WebAssembly(patter.io)
27 points by adrianbg 9y ago|5 comments
49.
by adrianbg 9y ago
I wonder if this / gradient descent would work with integers, and with mod 2^n as the nonlinearity.
50.
Grain Entrapment(en.wikipedia.org)
3 points by adrianbg 9y ago|1 comments
51.
by adrianbg 9y ago
The "Who did What" dataset seems much better in this respect: Passage: Britain’s decision on Thursday to drop extradition proceedings against Gen. Augusto Pinochet and allow him to return to Chile is understandably frustrating ..
52.
by adrianbg 10y ago
Is this a cult?
53.
by adrianbg 11y ago
Unions originated in response to maltreatment of workers. How would you solve that problem without them?
54.
by adrianbg 11y ago
Is it me or could they have done this way more easily by building some indexing and triggering functionality on top of Cassandra? Even two years ago when they started. Instead they built sharding, indexing, triggering and a Cassandra-like d
55.
by adrianbg 11y ago
Really interesting. I wonder how politicians would rate.
56.
by adrianbg 11y ago
Not efficient at the start is a little different than "we have no reason to think it will ever be more efficient".
57.
by adrianbg 11y ago
E2's have to be financed by citizens of your country of origin. http://startupimmigrationattorney.com/e2-visa-for-startups-p...
58.
by adrianbg 11y ago
Yeah. Their investment comes in the form of a salary. I don't know the details for H1b's specifically but I can tell you that they haven't had trouble meeting visa requirements for people in the program.
59.
by adrianbg 11y ago
http://unshackled.co/ You can transfer your H1b to them and work on your thing.
60.
by adrianbg 11y ago
D-Wave's computers work as intended and exhibit "computationally relevant quantum effects" but we still don't know whether the strategy they are using will ever yield practical advantages. Other groups (Google, Universit

More ›