v2.28.0

Introducing Audio Transcription and Provider Enhancements

Sreejan
SreejanAuthor

This release introduces a major new capability with our Speech-to-Text API and expands the feature set of several underlying providers, reinforcing our commitment to offering a comprehensive and powerful AI gateway.

New Features

3
  • Launched Audio Transcription API: Introduced a new /v1/audio/transcriptions endpoint for converting speech to text, integrated across multiple key providers.
  • Enabled Text-to-Speech (TTS) Capabilities: Added TTS support for a key provider, enabling high-quality audio generation directly from text inputs.
  • Expanded Embedding Generation: Integrated embedding generation support for another key provider, broadening our text vectorization capabilities.

Improvements

3
  • Enhanced Audio Billing Accuracy: Implemented a more robust audio duration calculation method to ensure precise, fair billing for all transcription services.
  • Increased API Performance and Plan Limits: Significantly increased the number of API workers and raised the daily request limits for paid plans to improve overall performance and user capacity.
  • System Maintenance: Performed various internal cleanups, including the removal of an unused provider and synchronization of model configurations to enhance platform stability and maintainability.