PRO
afcorson
Australia
Asked
How can I record audio from the microphone for an extended period, such as up to 10 minutes? I want to capture a person speaking, save the audio as an MP3 file on my hard drive, and then send it to an AI platform for a response. I have attempted to use the Microphone skill in Synthiam ARC, but it failed to record anything and does not save the audio to my hard drive.
Could you provide guidance on how to achieve this functionality?
Related Hardware EZ-B v4
Related Control
Microphone
Thanks for the update to the Chromium skill with the global content variable. This works well and allows me to retrieve and speak the text. The value of the Chromium content variable could get quite huge if the person speaks for 5 mins or so. I will have to try out the whole process by recording a live speaker and test out the whole process.
Yeah that's true. Another option would be to have a robot skill that records audio for a long time and writes to a file to send to your ai platform.
What's the process you are currently taking to generate the file and send it to the ai platform? Because maybe if i have some time i can think of a robot skill that could be useful for stuff like this - instead of using advanced bing speech recognition i guess. I think the biggest challenge you're up against is how to perform text to speech on a super long audio recording.
There seems to be numerous online audio to text converters out there - this one I tried was very good (https://transcribetotext.ai). But I need to record a 5 min speech and thoroughly test this site. I'll try and do this in the next day or so and report back. I'll send the text to https://deepai.org/chat/debate and see what it responds with.
I recorded a 6 min speech today, used transcribetotext.ai to convert to text, and sent the text to deepai.org/chat/debate for a reponse. Within a few seconds a meaningful response came back which my robot could speak. So the process is feasible. Just need to streamline the process and test it in a live environment.
I'm not sure if that can be automated. They don't have an API for performing that - they only seem to have vision for their API. here's a screenshot of what i saw on their website for the list of api's
You are right there. But at the rate AI is developing, I am sure someone will eventually provide an API for audio to text and debate responses.
Will the chat gpt not do a debate response? If you instruct the personality for it, it should.
then we just need to figure out the long recording analysis of speech to text from an audio file.
I haven't tried ChatGPT for a debate type response. I'll give it a go and compare the results.