Deepgram

Deepgram is a leading provider of automatic speech recognition (ASR) technology that uses deep learning to accurately transcribe and analyze audio data. The company's AI-powered platform offers both speech-to-text and text-to-speech capabilities, with a focus on high accuracy, low latency, and cost-effectiveness. Deepgram's ASR technology can handle complex audio environments and diverse accents, supporting transcription in over 36 languages as of March 2024. The platform is designed to operate at scale, allowing it to process large volumes of audio data efficiently.

Deepgram's technology incorporates advanced features such as speaker diarization, sentiment analysis, topic detection, and language detection. The company's audio-first approach allows it to capture nuances of intonation and inflection, setting it apart from traditional text-based natural language processing methods. As of November 2022, Deepgram had transcribed over 1 trillion words from raw audio data.

In March 2024, Deepgram launched Aura, a new real-time text-to-speech API that combines highly realistic voice models with low-latency processing. This technology enables developers to build conversational AI agents for applications such as customer service and interactive voice response systems. Deepgram's pricing for its text-to-speech service is competitive, with rates starting at USD 0.015 per 1,000 characters as of March 2024.

Key customers and partnerships

Deepgram's technology has been adopted by a diverse range of enterprises and startups. Notable customers include NASA, Spotify, Auth0, and Citi. The company's solutions cater to various industries, including customer support, media, education, and healthcare.

In the education sector, Deepgram's technology is used to assist in language learning and provide resources for students. In healthcare, the platform helps transcribe doctor-patient conversations for improved record-keeping and compliance. Media companies utilize Deepgram to generate subtitles and closed captions for videos in multiple languages.

Deepgram has also established partnerships to enhance its offerings. The company is part of the NVIDIA Inception program, which has allowed it to leverage NVIDIA GPUs to improve the speed and scalability of its platform. This collaboration has enabled Deepgram to process hundreds of audio files in seconds using NVIDIA V100 GPUs on the Amazon Web Services cloud.