Voice Recognition AI Expertise

[I__Dali__I]

Looking for some expertise in voice recognition ai for a potential art installation. Please reach out if you know anything in this area or someone who could potentially help. Many thanks!!

[Token]

That will be real tough without a reliable internet connection.

Voice AI is practically a commodity these days but all the smarts is in the cloud.

For example: Siri on the iPhone can only do basic music playback controls when not connected to the net.

Look up voice tech that is not cloud dependent if you choose to go down this path. Snips is an example of standalone voice AI.

[BBadger]

You can train basic stuff using machine learning using something like Tensorflow. There are examples and such. It'll take quite a bit of training data, or the data that is often freely available, but it'll help you a lot to have an idea of what you want the models to do before you dive into it.

[I__Dali__I]

Thanks for the leads! Definitely need a standalone solution. Snips in particular looks interesting, going to contact them.

[some seeing eye]

Should you need outside connectivity to a cloud solution, the BORG does too and has two somewhat redundant Internet connections for their own department operations, like ESD. Usually they are reliable, but sometimes they are not - in 2017 early week for example, due to human error.

Contact the Artery if you need to use Internet connectivity in your art project.

Some people also bring consumer satellite ground stations. Maybe you and the software you select can stand the lag. There are a variety of vendors searchable by satellite Internet, from handheld devices to small dishes which have to be aligned.

There was a new festival WiFi LTE hot spot vendor this year. That will likely not work as the week becomes congested.

There is a lot of technical detail regarding antenna sectors and their probability to become congested which can come into play with your placement.

[Lonesomebri]

a9AvK0K_460s.jpg

[maladroit]

Look up voice tech that is not cloud dependent if you choose to go down this path. Snips is an example of standalone voice AI.

I've experienced a Snips setup, it is what you want. They are designing it specifically for applications that won't or can't use a cloud-hosted service.

[Token]

Snips runs on a Raspberry Pi 3

‘Nuff said.

Perfect for the wilderness of the Playa.

[apophn]

You can run a voice recognition system on a Pi or a small computer easily, it doesn't have to be in the cloud. Depending on the kind of model you're running and what kind of hardware you have available it may not be real-time, but it's certainly doable. The tricky thing is either training your own model or finding one somewhere you can use. I can lend a hand if you're interested though.