Deepgram
Scalable Speech-to-Text, Text-to-Speech & Voice Agent APIs with simple, transparent billing.
Overview
Deepgram is a developer-first Voice AI platform offering highly accurate, scalable Speech-to-Text (STT), Text-to-Speech (TTS), and Voice Agent APIs. Built for real-time and batch workflows, Deepgram supports over 45 languages, advanced features like Speaker Diarization, Smart Formatting, and Audio Intelligence models, combined with true per-second billing and enterprise-grade security.
Key features
14 highlightsSpeech-to-Text
Text-to-Speech
Voice Agent API
Speaker Diarization
Smart Formatting
Keyterm Prompting
Automatic Language Detection
Sentiment Analysis
Topic Detection
Summarization
Intent Recognition
Redaction
Entity Detection
Self-hosted VPC deployments
Pricing
Pricing not confirmed
We haven't verified a structured pricing table from the vendor's website. Check the tool's official pricing page for the latest.
Pros
- Provides a generous $200 free credit to new accounts that does not expire after 12 months.
- Employs true per-second billing without rounding up to the nearest minute or 15 seconds.
- Charges the exact same low rate for both real-time streaming and pre-recorded (batch) transcription.
- SOC 2 Type 2 certified and HIPAA compliant, with BAAs available for Enterprise customers.
- Provides options to deploy on-premise or in private clouds via self-hosted containers.
- Offers a Growth plan starting at $4k/year that unlocks up to approximately 20% discount across products.
- Aura TTS Billing is done by input character, facilitating precise cost controls.
- Supports automatic multilingual transcription and code-switching via Nova's multilingual models.
- Allows refunds on unused purchased credits within 30 days of purchase.
Cons
- Self-hosted container deployments are restricted to Enterprise tier and require dedicated NVIDIA GPUs.
- Promotional or free credits are completely non-transferable under all circumstances.
- Purchased credits are non-refundable once they pass the 30-day mark.
- No refunds or adjustments are provided for incorrect or faulty audio files uploaded due to user-side system/coding errors.
- Exceeding currency limits on Pay-As-You-Go plans may cause requests to be queued or rejected.
- Multichannel audio billing scales by total processed duration (e.g., 2 channels for 10 minutes equals 20 minutes of billable time).
Platforms
Community reviews
Verified community feedback. Ratings are calculated from approved reviews only. How this works
No community reviews yet
Be the first to review this tool.
Related tools
More tools in Voice, based on category and tags.
Recently added
Fresh additions to the directory
