LibreASR
AI
speech recognition
SHOW HN2020
How HN took it
2 positive, 18 neutral or asking, 0 mixed, 2 critical, out of 22 comments about the project. Thread score +0.04.
Performance: 0 for, 2 againstIdea: 1 for, 0 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Good idea to run ESP32-LyraT sample and under-appreciated boards
From 1 comment
I LOVE that you provided a sample application targeting the ESP32-LyraT! While the ESP8266/ESP32 get plenty of love on HN (and elsewhere) I think the ESP ADF (audio development…
kkielhofner on HN
Objections
Transcription quality is poor compared to cloud services and YouTube captions
From 2 comments
As I commented above, very poorly. It's still early days.
iceychris on HNThe upper transcript is YouTube's automatic transcription. Below is the web app transcribing live. And yes, it is actually missing a few words.
iceychris on HN
RNN-T is very hard to train and a single GPU is insufficient
From 1 comment
The secret of RNN-T is that it is __extremely__ hard to train, it is very unstable. On a single GPU you'll spend years to train reasonable quality model, moreover a streaming…
nshm on HN
Requests
Add French and other language support
From 1 comment
Hi! What should you need to implement other language i.e. Italian or French? I mean: it's a problem due to the less of datas or what? Another question: could you use for example…
th3h4mm3r on HN
Document CPU and storage requirements
From 1 comment
The README could use a section as to what CPU platform and storage requirements are necessary to run this app.
donpdonp on HN
Report WER per dataset
From 1 comment
Can you post WER per dataset? Bucketing all of the WER together means you can only directly compare to models that are validated on the exact same combination datasets. This…
lunixbochs on HN
Questions
What datasets are used for training
From 1 comment
LibriSpeech, Tatoeba, Common Voice and scraped YouTube videos.
iceychris on HN
How does transcription quality compare to cloud offerings
From 1 comment
(Anyone know how the transcription quality compares to the various cloud offerings from AWS, Google, and IBM?)
saurik on HN
Alternatives named
Vosk
From 2 comments
Mozilla DeepSpeech has streaming audio support as of a few releases ago, the word error rate has also improved. I would recommend looking at Vosk too, it converts speech to text…
posguy on HNCheck out Kaldi. It's a toolkit rather than a ready-to-deploy service but has some solid pretrained models and recipes for training your own. You can use various existing…
woodson on HN
wav2letter
From 2 comments
Use wav2letter
whimsicalism on HN> open-source or paid SDK/API that I can use to create a group voice chat mobile app with "live" transcription? Or something that can plug-in to a system like this? Yes, Google,…
whimsicalism on HN
Mozilla DeepSpeech
From 1 comment
Mozilla DeepSpeech has streaming audio support as of a few releases ago, the word error rate has also improved. I would recommend looking at Vosk too, it converts speech to text…
posguy on HN
Kaldi
From 1 comment
Check out Kaldi. It's a toolkit rather than a ready-to-deploy service but has some solid pretrained models and recipes for training your own. You can use various existing…
woodson on HN