Arize Release Notes: Voice Application Tracing and Evaluation

What’s New

Voice Application Tracing and Evaluation

Capture, process, and send audio data to Arize. Instrument your audio application to send events and traces to Arize, capture key events from the OpenAI Realtime API’s WebSocket and convert them into spans that provide meaningful insights into your system’s behavior. Assess how well your models identify emotional tones like frustration, joy, or neutrality to enhance user interactions with evaluation.

Key features include:

• Audio Tracing: Gain visibility into the entire workflow of voice applications, from input to output, to identify bottlenecks or errors.
• Evaluation: Assess how well your models identify emotional tones like frustration, joy, or neutrality to enhance user interactions.

Enhancements

Dashboard Updates

We’ve made several enhancements and fixes to our dashboards to improve usability and performance.

New Functionality

• Hide Overall Metric for Group By: Users can now choose to hide the overall metric when using group by functionality for a cleaner view.
• Custom Color Support: Personalize your dashboards with custom color options for better visual distinction.

Miscellaneous Improvements

• Legend Fixes: Resolved issues with legend display for better clarity and consistency
• Improved Axis and Legend Handling: Enhanced support for long axis and legend labels to maintain readability
• Dashboard Creation Modal Cleanup: Simplified the creation modal by reducing the number of required fields for a smoother user experience

📚 New Content

The latest video tutorials, paper readings, ebooks, self-guided learning modules, and technical posts:

🛠️ Building an Agent Router: Best Practices
🧑‍⚖️ LLMs as Judges: A Comprehensive Survey on LLM-Based Evaluation Methods
🤔 Understanding Agentic RAG
🇪🇺 Quick Guide to the EU AI Act
👾 Training LLMs to Reason in Continuous Latent Space
🚚 How Geotab and Arize Revolutionized Fleet Management