Skip to content

AudioBard

Open-source multi-voice audiobook generator from EPUB and TXT books.
Transform books into cast-narrated audiobooks with distinct, consistent voices assigned to each character — completely offline or via BYOK cloud models.

AudioBard Banner

Motivation

Standard text-to-speech tools read entire books in a single robotic monotone voice without character distinction. Commercial voice platforms frequently require uploading private manuscripts to remote servers and charge recurring subscriptions.

AudioBard provides a local-first alternative that automatically extracts character rosters, attributes dialogue with emotional context, and synthesizes audio tracks using neural TTS models.

AudioBard Pipeline

Key Capabilities

  • Character Extraction & Dialogue Attribution: Uses LLMs to detect speaking characters, track aliases across chapters, and determine vocal emotion.
  • Multi-Voice Neural Synthesis: Automatically maps distinct voices (via Piper TTS or Edge TTS) to each character based on gender, age, and emotional tone.
  • Offline & Cloud Modes: Complete local privacy with Ollama + Piper TTS (zero data leaves your machine), or cloud BYOK (NVIDIA NIM, OpenRouter, Google Gemini, Edge TTS) with your own API keys.
  • Desktop GUI & CLI: Built with Tauri v2 + Vue 3, featuring drag & drop book ingestion, library player, and Spanish / English localization.

Comparison with Alternatives

Feature AudioBard Commercial Cloud Platforms Standard Reader TTS
Execution Environment Local offline or Cloud BYOK Cloud-hosted servers Local device
Character Casting Multi-voice per character Manual studio configuration Single narrator voice
Interfaces Desktop GUI (Tauri) + CLI Web dashboard only Desktop or browser extension
Licensing & Cost Open source (MIT), zero subscription fees Subscription ($15–$100+/mo) Free or bundled
Attribution Verification Hermetic benchmark suite vs gold standard Not published Not applicable

Documentation Navigation