​
PodCraft: AI-Powered Conversational Audio

Unleash the power of conversational AI with PodCraft! Generate expressive, long-form, multi-speaker audio, such as podcasts, from text.

​
Overview

PodCraft is a novel framework designed for generating expressive, long-form, multi-speaker conversational audio, such as podcasts, from text. It addresses significant challenges in traditional Text-to-Speech (TTS) systems, particularly in scalability, speaker consistency, and natural turn-taking. A core innovation of PodCraft is its use of continuous speech tokenizers (Acoustic and Semantic) operating at an ultra-low frame rate of 7.5 Hz. These tokenizers efficiently preserve audio fidelity while significantly boosting computational efficiency for processing long sequences. PodCraft employs a next-token diffusion framework, leveraging a Large Language Model (LLM) to understand textual context and dialogue flow, and a diffusion head to generate high-fidelity acoustic details. The model can synthesize speech up to 90 minutes long with up to 4 distinct speakers, surpassing the typical 1-2 speaker limits of many prior models.

​
How to Use PodCraft

1

Compose a Script

Compose a script. The specific format for the script should be as follows:“Speaker 0: Hello there and today we are talking about this awesome new tool in Pixio called PodCraftSpeaker 1: Podcraft lets you generate podcasts with up to 4 speakersSpeaker 0: And it also allows you to upload custom voices.Speaker 1: Its really fast too!”You must start with Speaker 0. You can have up to 4 speakers.
Compose a Script
2

Add Speakers

In the section below, add speakers from existing voices or select custom and upload an mp3 of a voice to reproduce.
Add Speakers
3

Finetune Options

Finetune the options for Generation speed and CFG scale. Slower speed produces higher quality results.
Finetune Options
4

Generate

Click generate. Depending on the length of the script, PodCraft can take up to 5 minutes.