Conformer-2

Introducing Conformer-2, the highly advanced AI-powered speech recognition model from AssemblyAI. Built upon the cutting-edge performance of its predecessor, Conformer-1, this state-of-the-art model has been rigorously trained on an expansive 1.1 million hours of English audio data to deliver unparalleled accuracy. With substantial enhancements in recognizing proper nouns, alphanumeric characters, and noise resilience, Conformer-2 is engineered to thrive in the challenging audio environments of today's diverse applications. This remarkable tool is poised to transform the way developers and businesses interact with voice data, offering reduced transcription latencies and robust performance in real-world scenarios. Discover the future of speech recognition with Conformer-2, now available for seamless integration through our API.

Key Features

• Extensive Training: Trained on 1.1 million hours of English audio data for unmatched accuracy.

• Enhancements in Recognition: Significant improvements in recognizing proper nouns and alphanumeric sequences.

• Superior Noise Robustness: Advanced robustness to noise ensuring clear transcriptions even in noisy environments.

• Speed Optimizations: Reduced inference latency making Conformer-2 faster than its predecessor.

• Accessibility through API: Easy integration and accessibility through a well-documented API for developers.

FAQ

What is Conformer-2?

Conformer-2 is an AI model for automatic speech recognition that has been trained on 1.1 million hours of English audio data.

What are the improvements in Conformer-2 over Conformer-1?

Conformer-2 features improvements in recognizing proper nouns, alphanumeric characters, and is more robust to noise compared to its predecessor.

How can I try out or integrate Conformer-2?

You can try out Conformer-2 through the API Playground or integrate it into your application using the provided API.

How does Conformer-2 compare with Conformer-1 in terms of processing speed?

Conformer-2 processes speech more quickly, with up to 53.7% decreased latency in transcription compared to Conformer-1.

Is Conformer-2 now the default speech recognition model on AssemblyAI's API?

Conformer-2 is now available as the default model through AssemblyAI's API, providing better performance with no need for changes from current users.