Main Purpose
AssemblyAI is an AI-powered speech recognition platform that provides accurate speech-to-text transcription for various applications.
Key Features
- Conformer-2: AssemblyAI's state-of-the-art speech recognition model trained on 1.1 million hours of data.
- Model Ensembling: Utilizes multiple strong teacher models to produce more robust student models.
- Scaling up to 1M+ hours: Increased model size to 450M parameters and trained on 1.1 million hours of audio data.
- Speed Improvements: Conformer-2 is faster than Conformer-1 by up to 55% depending on audio file duration.
Use Case
- Generative AI applications leveraging spoken data.
- Real-world use cases requiring accurate speech-to-text transcription.
