Meta has released an “open” implementation of the viral generate-a-podcast tool in Google NotebookLM.
The NotebookLlama project, as expected, makes extensive use of Meta’s own Llama models for processing.
It, like NotebookLM, can generate back-and-forth, podcast-style digests of text files that have been uploaded.
Meta has spent time perfecting an AI component that enabled Google NotebookLM to go viral.
To assist users in absorbing difficult information, Meta is offering an open implementation model of the generate-a-podcast capability.
It was launched under the name NotebookLlama and shares much of the same features, limitations, and potential as Google’s NotebookLM.
NotebookLlama is an AI-powered podcast generator that allows users to submit a PDF file and convert it into an audio podcast featuring two AI characters.
The tool completes the procedure using three different Llama 3.1 AI models.
NotebookLlama’s podcast, like Google’s, is based on a free-flowing conversation between two artificial intelligence hosts.
According to materials released on GitHub, the solution was created using Meta’s Llama model and is now available for public use.
NotebookLlama takes user-uploaded PDF files through a series of phases that employ Meta’s Llama 3.1 models. Initially, the Llama 3.2 1B model translates the PDF to text format.
The Llama 3.1 70B model then creates a podcast script based on the text, while the smaller Llama 3.1 8B model adds a conversational tone.
Meta’s Parler TTS technology also converts the script to audio, resulting in a dynamic dialogue between the AI characters.
We earlier reported that Meta Platforms Inc. has signed a licensing agreement with Reuters, granting it access to the news agency’s content.
