VoiceGPT is an app that has been developed specifically for those with motor disabilities.
It allows the user to ask for any information by speaking their queries to the AI and VoiceGPT converts this speech to text,
renders the request, fetches the required data, prepares a response, and sends it to the user in text and speech format.
VoiceGPT uses speech recognition to recognize user requests and responses.
It's predecessor, ChatGPT is an easy-to-use chatbot that allows users to streamline their workflow by gaining access to worldwide knowledge through text communication.
However, this poses a problem for those with physical disabilities who can't physically type their problems. Our advanced version of ChatGPT focuses on text-to-speech and speech-to text communication, similar to voice assistants like Siri, Alexa and Bixby. This app makes life easier for those with disabilities as well as normal folks, you don't need to waste time typing anymore!
Firstly, we had to make our own custom API, which is quite a challenge for beginners. Our API bypasses the rate limitations and API purchases of the original ChatGPT by OpenAI. The rerouting and precise usage of session tokens posed a hurdle in building the backend of the API.
To solve this, we took the email credentials from the user, creating session tokens and using them as bypass authenticators in the beta version of ChatGPT. Using this loophole, we can use the latest API's; OpenAI v3.5 Turbo on which ChatGPT and VoiceGPT are based. This allowed us to use the vast range of ChatGPT combined with the ease of accessibility provided by voice recognition technology, literally the best of both worlds!
Tracks Applied (3)
Technologies used
Discussion