Whisper is an open source model that was developed by OpenAI, and automatically recognizes and transcribes human speech to text. The model is capable of transcribing in multiple languages (our team has tried Spanish, Portuguese, Japanese, and Urdu with great success!) as well as translating from those languages into English. In this video, Justin will show you how to access the Whisper model API on Baseten to build a simple yet elegant speech-to-text translation app.
You can follow along and build your own, or check out our Whisper API app here: https://app.baseten.co/apps/b0dgKEB/operator_views/mP7AXaB
You can get started on Baseten today by signing up at https://www.baseten.co/. There’s no credit card needed for our free tier!
Join The Data Science Community: https://community.baseten.co/
Socials:
YouTube – https://www.youtube.com/channel/UCOCLmqf7Jy3LcsO0SMBGP_Q
Twitter – https://twitter.com/basetenco
LinkedIn – https://www.linkedin.com/company/baseten
Sources:
Baseten – https://www.baseten.co/
Truss – https://truss.baseten.co/
Whisper source code – https://github.com/openai/whisper
Launch announcement – https://openai.com/blog/whisper/
This video was filmed and edited by Jesse Mostipak (@kierisi)
#whisperapi #openaiwhisper
Meet Blueprint, our latest product offering: https://blueprint.baseten.co/
Blueprint allows you to fine-tun and integrate generative models into your APIs, and is the fastest way for you – the developer- to customize ML models. Start your project with pre-configured environments, GPUs, API endpoints, and instant deployment.
With Blueprint you can run generative models locally. No GPUs required.
baseten.models (now on PyPi!) is a Python package that gives you full customizability over models like Stable Diffusion, Whisper, and FLAN-T5 from your local machine. Go beyond standard web API limitations to build the next generation of ML-powered applications.
Blueprint also gives you easy-to-use dev tools, inspired by the latest in AI research.
There’s never been a more exciting time in machine learning than today. New, state of the art research is published daily. We’re translating these findings into easy-to-use developer tools to unlock new ways for developers to build with ML.
