# Deepgram Nova-2
> Deepgram Nova-2 is a proprietary speech-to-text (STT) model developed by Deepgram. It utilizes a specialized transformer-based architecture designed to optimize for both transcription accuracy and processing speed. The model is primarily delivered via API for integration into enterprise software, meeting applications, and contact center technologies.
- URL: https://optimly.ai/brand/deepgram-nova-2
- Slug: deepgram-nova-2
- BAI Score: 68/100
- Archetype: Challenger
- Category: Artificial Intelligence
- Last Analyzed: March 12, 2026
- Part of: Deepgram (https://optimly.ai/brand/deepgram)
## Competitors
- AssemblyAI (https://optimly.ai/brand/assemblyai)
- Google Cloud Speech-to-Text (https://optimly.ai/brand/google-cloud-speech-to-text)
- Openai Whisper (https://optimly.ai/brand/openai-whisper)
## Buyer Intent Signals
Problems: Manual Transcription: Transcribing audio files by hand or using internal staff to listen and type out content. | Human Transcription Agencies: Outsourcing large batches of audio to human transcription firms like Rev.com.
Solutions: fastest speech to text API | lowest WER transcription model | real-time voice to text for developers | best free transcription tool for students | Deepgram Nova-2 benchmarks | Self-hosted Open Source (Whisper): Utilizing open-source models like OpenAI's Whisper but managing the infrastructure, GPU hosting, and scaling internally.