rayraylalonde-sys/ab_studio

AB_Studio is a two step Windows tool for turning a written book into a finished audiobook using Microsoft Azure's text to speech service, specifically its Dragon HD Neural voices. It works on ODT, EPUB, or plain text files, and the whole process runs as standalone Windows programs, so no Python installation is needed just to use them. The first stage is a program called TTS Optimizer. It takes a manuscript and splits it into one text file per chapter, while also cleaning up the text in ways that prevent common speech synthesis problems: it removes image references, converts dashes to commas since the voice engine reads them poorly, strips section dividers, removes back matter like tables of contents, fixes stray capitalized words, and smooths out punctuation spacing issues that would otherwise cause audible stutters. A user can also list specific character names so the tool applies a comma fix around them for more natural sounding dialogue. The second stage is AB Studio itself, which sends the cleaned chapter files to Azure and saves the results as WAV audio files, automatically splitting and stitching together long texts as needed. A user picks a voice, adjusts speaking rate, pitch, and pause lengths, chooses an output format, and can preview a short clip before rendering the full file. Each output WAV file is ready to use directly in an audiobook project. Using this requires a free or paid Microsoft Azure Speech Services account, since the actual voice generation happens through Azure rather than locally. The free tier includes 500,000 characters of speech per month, and the tool itself does not track how much of that has been used, so a user needs to check their Azure account for that. Processing ODT or EPUB files also requires installing a separate free tool called Pandoc. Building the programs from source requires Python 3.10 or newer. The project is released under the MIT license, so it can be used freely, including for commercial purposes.