Audio & Speech

Today's Stories

1. 1

   ElevenLabs launches AI that creates songs blending opera and heavy metal seamlessly

   ElevenLabs released Music v2 on May 28, an AI music generator that can transition between completely different genres within a single song—from opera to heavy metal to rap without losing musical coherence. The new version includes an 'inpainting' feature that lets users regenerate specific sections of a song without affecting the rest.

   Musicians and content creators can now experiment with genre-blending compositions that would be extremely difficult to create manually, opening up new creative possibilities for film scores, video games, and experimental music.

2. 2

   Spotify launches AI audiobook creator powered by ElevenLabs voice technology

   Spotify introduced an audiobook creation tool on May 21 that uses ElevenLabs' AI voice technology to automatically convert written books into spoken audio. Authors can use the service without signing exclusive contracts, meaning they can publish their AI-generated audiobooks on other platforms as well.

   Authors who couldn't afford professional narrators can now create audiobooks at a fraction of the cost, potentially flooding the audiobook market with new titles and making audiobooks more accessible to readers.

3. 3

   Open-source speech recognition models challenge expensive cloud services

   Several new open-source speech-to-text models emerged, including MOSS-TTS v1.5 with improved voice cloning and multilingual support. Developers are seeking self-hosted alternatives to expensive services like AssemblyAI, with some creating custom pipelines using multiple AI models and audio processing techniques to improve accuracy.

   Small businesses and developers can now access high-quality speech recognition without paying monthly fees to cloud providers, making voice-controlled apps and transcription services more affordable to build.

4. 4

   AI video avatars get major upgrade for realistic lip-sync and body movements

   LongCat-Video-Avatar 1.5 was released on May 23, featuring improved audio processing using Whisper-Large technology for more natural lip synchronization. The open-source framework can create realistic talking avatars from audio input, with better stability for long videos and support for different visual styles including anime.

   Content creators, educators, and businesses can generate professional-looking video presentations with virtual spokespersons, reducing the need for human actors or expensive video production.

5. 5

   Voice AI companies shift from selling tools to delivering complete services

   AI service providers are moving away from selling individual voice AI components like chatbots or phone systems, instead offering complete business solutions. Rather than pitching 'voice AI for $500/month,' companies now sell 'complete customer service management for $4,000/month' while quietly using AI agents behind the scenes.

   Businesses will get more comprehensive solutions for customer service and sales, but may pay higher prices as AI companies package their technology into full-service offerings rather than standalone tools.