DeepSpeech

You are right, and I expect Reuben and team will fix this and are grateful for the heads up. I was able to find the right links though, for example the WAV transcriber is at : https://github.com/mozilla/DeepSpeech/blob/v0.6.0/examples/v...

It's using letters, not words, so in theory anything that can be spelled in English can be output: https://2r4s9p1yi1fa2jd7j43zph8r-wpengine.netdna-ssl.com/fil... Or, if you prefer the code: https://github.com/mozilla/DeepSpeech/blob/0427c1572ac8f7253... (well, I suppose technically it's in https://github.com/mozilla/DeepSpeech/blob/master/data/alpha... through https://github.com/mozilla/DeepSpeech/blob/master/util/text...., but I mean come on) In practice ... meh. It gets some things predictably wrong. They're readable, just wrong. So the good news is that WER is words it has 100% correct. The remaining errors aren't usually totally unreadable, just irritating, going from "totext" instead of "to text", but of course names go pretty wrong and don't get better. It butchers my name, of course. If you test it yourself, please do realize it r

There's a basic example here transcribing from an audio file: https://github.com/mozilla/DeepSpeech/tree/v0.6.0/#project-d... There are also examples transcribing from the microphone as I mentioned above: https://github.com/mozilla/DeepSpeech/tree/master/examples/ I also have a small GUI example here [0], click once to start recording, once again to stop and show transcript. It receives the same command line arguments as the ones shown in the README link above (namely --model, --lm and --trie). The README starts with a very basic installation and usage guide, I don't see how that's assuming any expertise. [0] https://github.com/reuben/ds-qttest

> 1. I want to teach it ten words. How do I do this? I'm assuming you mean "my target vocabulary is only ten words". In that case, you have a couple of options: 1. Just use it as is and see how it does. 2. Create a language model containing just those 10 words so that the engine is constrained to that vocabulary, which should improve accuracy. 3. Train a (smaller) model from scratch, which requires (audio, transcript) pairs. This is basically research work, there's no recipe you can just follow. The advantage is that if you can make it work, it can be a much smaller model than a model trained for general English dictation. > 2. I want to speak into my microphone (available as a Pulseaudio device) and recognise the words and output the words as a text stream on stdout. How do I do this? We have a few examples with microphone input, but th

In the post I mention a Brazilian company who's using DeepSpeech for vocabulary constrained, medical applications. They've trained their own Brazilian Portuguese models. The English acoustic model that we released does not have a fixed vocabulary. What determines the vocabulary is the language model, which can be created from text. So if you have in-domain text, and would like to try it out, I would say the first step is to create a language model using your text and then experiment with it. We have some documentation on how to create the LM here: https://github.com/mozilla/DeepSpeech/tree/v0.6.0/data/lm It's not super detailed, but I'd be happy to answer questions on our Discourse: https://discourse.mozilla.org/c/deep-speech

> how many GB are we talking to be able to cover normal daily speech The models of Mozilla's DeepSpeech STT engine take 1.8 GB in compressed form: https://github.com/mozilla/DeepSpeech/releases/tag/v0.5.1 The main cause for the large size is the language model. They tried using different (smaller) language models, but they weren't as good.

Mozilla has done some work on text-to-speech and the other way around too (I think). [https://github.com/mozilla/DeepSpeech] Not so surprising that they want to stay on the cutting edge. The best way to shape the future is too invent it (misquoting Alan Kay)

Mozilla is maintaining an implementation of Baidu's deepspeech with pretrained models (and checkpoints so you can fine tune with your own dataset). They have an example that accepts streaming from the microphone: https://github.com/mozilla/DeepSpeech/tree/master/examples/m... See the last full release here: https://github.com/mozilla/DeepSpeech/releases/tag/v0.5.1

Maybe try Mozilla DeepSpeech. https://github.com/mozilla/DeepSpeech

I really appreciated this article as it presents the kinds of challenges that all sorts of people can find them with unexpectedly, and shows how perseverance and the help of technology can be a way round them. More specifically I was interested to see that the version of Dragon for scripting isn't available any longer. However this might be something where Mozilla's recent efforts with speech recognition could help and as a bonus the project is open-source and adaptable. https://github.com/mozilla/DeepSpeech The WER they've achieved is already good for general recognition and through experimenting with narrowed domains / language models it can be made extremely accurate. For anyone interested, there's a walkthrough I did of steps needed to set up a narrow domain LM on their forum below with a video showing it working. I'm guessing that a

I dunno, but here's a way to visual voicemail your own mail (this is how I have been doing it since 2015): 1. sign up with a voip provider that either terminates on your own pbx (e.g. asterisk/freepbx) or provides voicemail with emailed audio files. setup said service with voicemail. 2. setup your favorite container system and let systemd manage the service 3. create a container that watches for new audio voicemails , when it finds a new voicemail push it to your favorite cloud provider (s3, gcp bucket, azure etc) then trigger that providers speech-to-text function on the audio file in the related cloud storage. store the resulting text message and metadata in a db. 4. create another container that simply watches that db for new messages and push out alerts however you see fit. you have many paths here. i went with a web-based ui that p

I think there are a few companies running Mozilla DeepSpeech in production: https://github.com/mozilla/DeepSpeech IIRC its mostly TensorFlow & Python though.

mozilla has open source projects for both speech-to-text and text-to-speech, among other things https://github.com/mozilla/DeepSpeech https://voice.mozilla.org/en

What's your metric? And are these offline assistants smartphone apps? NLP is expensive computing, so it'd make sense for those to do poorly. And this doesn't even have to strictly be offline; You can have your data processed online in a privacy-respecting way, so long as you actually know what's going on under the hood and have control over the data. My shill is MyCroft[0]. They're partnered with Mozilla to develop an open source NLP engine[1], so I have high hopes for it. [0] https://mycroft.ai/ [1] https://github.com/mozilla/DeepSpeech

I think the most interesting part of this is the offline capability. From looking at the source code, it looks like the text to speech is provided by http://www.festvox.org/flite/ and the speech to text uses https://github.com/mozilla/DeepSpeech. How do those compare to the major online providers?