Deepgram

Deepgram provides speech AI APIs for developers and enterprises. Its products cover speech to text transcription, text to speech synthesis and a voice agent API for real time conversational applications. Audio intelligence features such as speaker diarization, smart formatting, redaction, entity detection and keyterm prompting can be added to transcripts. Its recognition models are hosted as APIs, and enterprise customers can request custom trained models and deployment options. The platform is aimed at developers, startups and businesses building voice agents, transcription pipelines and audio analysis tools. The capabilities that matter most are streaming latency, turn detection for voice agents, multilingual recognition with support for more than 45 languages in its Nova models, and the choice between prerecorded and streaming processing. Pricing is pay as you go with no minimum commitment, and new accounts receive a $200 credit before paid usage starts. A Growth tier with prepaid credits for the year saves up to 20 percent. Regular pay as you go rates include $0.0077 per minute for streaming monolingual speech recognition, $0.0092 per minute for streaming multilingual recognition and $0.0043 per minute for prerecorded monolingual recognition. Growth rates are lower. The vendor also shows temporary promotional streaming rates on its pricing page, which are not quoted here. Enterprise pricing is by quotation.