More Voices, Better Audio, and AI Cover Art

This week we expanded voice options with all 29 Google Gemini voices, fixed critical audio bugs, and shipped an AI-powered cover generation system with 16 visual styles to choose from.

We've been busy this week improving the audio generation pipeline and expanding your creative options. The highlight? You now have access to Google's full catalog of natural-sounding voices, plus a brand new way to create professional podcast cover art.

New Features

AI-Powered Cover Generation

Generate professional podcast cover art in seconds using AI. Choose from 16 carefully designed visual styles ranging from minimalist to cinematic, then let our system create custom artwork that matches your podcast's tone and content. The system runs in the background while you work on other things, and you'll get notified when your cover is ready.

Full Google Voice Library

We've expanded from a handful of curated voices to all 29 official Google Gemini voices. Each voice has been categorized and documented so you can find the perfect match for your podcast's personality. Whether you need a warm conversational tone or a more authoritative presence, there's a voice that fits.

Automated Weekly Changelog

You're reading it right now. We've set up an automated system that generates these updates every week, so you'll always know what's new and what's been improved.

Improvements

Smarter Script Processing

The text-to-speech engine now automatically removes section markers like [INTRO] and [OUTRO] before generating audio, so they won't be read aloud during your episodes. We've also added better handling of emotion tags to ensure they're interpreted correctly by the voice engine.

Better Audio Assembly

Episodes now include your custom intro and outro jingles in the correct order, creating a more polished listening experience. The audio pipeline has been updated to match the expected flow from branded intro through content segments to branded outro.

Enhanced Voice Previews

When selecting voices, the preview system now uses the latest voice models for more accurate representation of how your final audio will sound.

Knowledge Base Isolation

Your podcast's knowledge base is now properly isolated from other podcasts in your account, preventing knowledge from one show appearing in another's content suggestions.

Bug Fixes

Audio Duration Fixed

Resolved an issue where some episodes would play audio for only a portion of their total duration, with the rest being silence. The fix ensures all generated audio plays correctly from start to finish.

Production Stability Improvements

Fixed several issues that were causing errors in the admin interface, including missing voice options and database connection problems.

Voice Selection Reliability

Corrected an issue where selecting certain voices would fail during generation, ensuring all available voices work correctly.