Layla 5.5.0 has been published!

This release brings exciting new features that our community has been requesting, along with important improvements and bug fixes that enhance the overall user experience.

Layla's UI is now translated in multiple languages:

• English

• Chinese (中文)

• Korean (한국어)

• Polish (Polski)

• Russian (Русский)

Introducing Offline Transcription

Ever needed to transcribe a lecture, audiobook, or video but didn't have a reliable internet connection? Our new Offline Transcription feature has you covered. This powerful addition allows you to:

• Transcribe audio content in real-time without any internet connection

• Convert lectures into text for easier studying and note-taking

• Transform audiobooks into readable format

• Extract text from videos for accessibility or reference

This feature works entirely on your device, ensuring your privacy while providing professional-grade transcription capabilities wherever you are.

Bringing Characters to Life

We've also given our preset characters a visual upgrade! All preset characters now feature animated backgrounds that add depth and personality to your conversations. These dynamic backgrounds create a more immersive and engaging experience, making interactions with your AI companions feel more vibrant and alive.

Enhanced Speech-to-Text Customisation

You now have more control over speech recognition with the ability to configure:

• Minimum silence length detection

• Other advanced speech-to-text settings

• Fine-tune the system to match your speaking style and preferences

• Faster Response Control

The "stop generating" function is now more responsive, giving you better control over AI responses when you need to interrupt or redirect the conversation.

Smoother UI Performance

We've optimised how responses are generated and displayed, resulting in:

• More consistent response speeds throughout the entire message

• Elimination of the frustrating slow-start, fast-finish pattern

• Overall smoother text generation experience

Advanced Settings Addition

Power users will appreciate the new TopK parameter in advanced settings, providing even more fine-grained control over AI response generation.

Critical Bug Fixes

This update also addresses several important bugs that were affecting user experience:

• DuckDuckGo News Search: Fixed the issue preventing news searches from working properly

• Response Coherence: Resolved the bug where multiple regenerations would cause increasingly incoherent responses

• Text-to-Speech Flow: Eliminated unnecessary pauses after the first sentence during text-to-speech playback

Full changelog

New features

• app UI has been translated to multiple languages! Currently supports: English, Chinese, Korean, Polish, Russian

• Offline Transcription: transcribe audio such as lectures, audiobooks, or videos in real time

• all preset characters now have animated backgrounds!

Improvements:

• allow configuring minimum silence length and other settings for speech-to-text

• made stop generating slightly more responsive

• improved UI performance when generating responses, should reduce speed inconsistencies where response times sometimes start slow and speeds up at the end of the message

• added TopK in advanced settings

Bug fixes:

• fixed bug where DuckDuckGo news search was not working

• fixed bug where multiple regens make the model increasing incoherent

• fixed bug where text-to-speech sometimes pauses unnecessarily after the first sentence