Deepgram
Voice recognition software
Text to speech software
Deep learning software
Generative AI software
Synthetic media software
- Features
- Ease of use
- Ease of management
- Quality of support
- Affordability
- Market presence
Take the quiz to check if Deepgram and its alternatives fit your requirements.
Pay-as-you-go
Free Trial
Free version unavailable
Small
Medium
Large
- Transportation and logistics
- Media and communications
- Arts, entertainment, and recreation
What is Deepgram
Deepgram is an API-first speech AI platform that provides automatic speech recognition (ASR) and related audio intelligence capabilities for developers and product teams. It is commonly used to transcribe calls, meetings, media, and real-time voice streams, and to build voice-enabled applications. The product emphasizes programmatic access, streaming transcription, and model options that can be tuned for specific domains and vocabularies. It also offers text-to-speech capabilities for generating synthetic voice output in applications.
API-first for developers
Deepgram is delivered primarily as APIs and SDKs, which fits teams embedding speech capabilities into applications rather than using a standalone end-user app. It supports both batch and real-time streaming workflows, which is important for contact center, live captioning, and voice agent scenarios. The platform design aligns with engineering-led evaluation and integration patterns common in this category.
Real-time streaming transcription
The product supports low-latency streaming ASR, enabling live captions, agent assist, and conversational interfaces. Streaming support reduces the need for customers to build their own chunking, buffering, and partial-result handling logic. This is a practical differentiator versus solutions that focus mainly on post-call or file-based transcription.
Domain customization options
Deepgram provides features intended to improve accuracy for specialized terminology, such as custom vocabulary/keywords and configuration controls. These capabilities help teams adapt recognition behavior to industry jargon, proper nouns, and product names. For organizations with repeatable audio domains (e.g., support calls or regulated workflows), this can reduce manual correction effort compared with generic-only models.
Requires engineering integration effort
Deepgram is primarily a developer platform, so value realization depends on integration work, monitoring, and ongoing tuning. Teams seeking a turnkey transcription workspace, note-taking UI, or end-user productivity tool may need additional software. Procurement and governance often require internal evaluation of API usage, logging, and data handling practices.
Accuracy varies by audio conditions
As with other ASR providers, performance depends heavily on audio quality, accents, background noise, and overlapping speech. No single model configuration fits all environments, so teams may need to test multiple settings and iterate. In noisy or multi-speaker scenarios, additional preprocessing or diarization tuning may be required to meet quality targets.
Cost scales with usage volume
API-based pricing typically scales with minutes processed and feature usage (e.g., streaming, diarization, or add-on capabilities). High-volume deployments such as large contact centers or media archives can see costs increase quickly without careful usage controls. Organizations often need budgeting guardrails, caching strategies, and selective feature enablement to manage spend.
Plan & Pricing
Pricing model: Pay-as-you-go with optional Growth (prepaid credits) and Enterprise (custom)
Free tier/trial: New accounts receive $200 of free credits at signup (no credit card required); after credits are used, billing is pay-as-you-go. Growth plan is pre-paid (annual) and offers up to ~20% savings; Enterprise is custom-priced.
Example costs (from official Deepgram pricing page):
-
Voice Agent API (per minute):
- Standard: $0.0800/min (Pay As You Go) | $0.0700/min (Growth)
- Standard - BYO TTS: $0.0600/min (PAYG) | $0.0500/min (Growth)
- Custom - BYO LLM + TTS: $0.0500/min (PAYG) | $0.0400/min (Growth)
- Advanced: $0.1600/min (PAYG) | $0.1500/min (Growth)
- Advanced - BYO TTS: $0.1200/min (PAYG) | $0.1100/min (Growth).
-
Speech-to-Text (streaming, per minute):
- Flux: $0.0077/min (PAYG) | $0.0065/min (Growth)
- Nova-3 (Monolingual): $0.0077/min (PAYG) | $0.0065/min (Growth)
- Nova-3 (Multilingual): $0.0092/min (PAYG) | $0.0078/min (Growth)
- Nova-1 & 2: $0.0058/min (PAYG) | $0.0047/min (Growth)
- Enhanced: $0.0165/min (PAYG) | $0.0136/min (Growth)
- Base: $0.0145/min (PAYG) | $0.0105/min (Growth)
- Add-ons (example): Redaction $0.0020/min (PAYG) | $0.0017/min (Growth); Keyterm Prompting $0.0013/min (PAYG) | $0.0012/min (Growth); Speaker Diarization $0.0020/min (PAYG) | $0.0017/min (Growth).
-
Speech-to-Text (pre-recorded/batch, per minute):
- Nova-3 (Monolingual): $0.0043/min (PAYG) | $0.0036/min (Growth)
- Nova-3 (Multilingual): $0.0052/min (PAYG) | $0.0043/min (Growth)
- Nova-1 & 2: $0.0043/min (PAYG) | $0.0035/min (Growth)
- Enhanced: $0.0145/min (PAYG) | $0.0115/min (Growth)
- Base: $0.0125/min (PAYG) | $0.0095/min (Growth)
- Add-ons (example): Redaction $0.0020/min (PAYG) | $0.0017/min (Growth); Entity Detection $0.0017/min (PAYG) | $0.0017/min (Growth); Keyterm Prompting $0.0013/min (PAYG) | $0.0012/min (Growth).
-
Text-to-Speech (billed per characters):
- Aura-2: $0.030 / 1k characters (PAYG) | $0.027 / 1k characters (Growth)
- Aura-1: $0.0150 / 1k characters (PAYG) | $0.0135 / 1k characters (Growth).
-
Audio Intelligence (task-specific language models):
- Summarization: $0.0003 / 1k input tokens and $0.0006 / 1k output tokens (PAYG) | $0.00024 / 1k input tokens and $0.00048 / 1k output tokens (Growth).
- Other audio-intelligence task pricing referenced on the pricing page; see official page for details.
Growth plan: Minimum commitment noted on the pricing page is $4k+ / year (prepaid credits) and offers up to ~20% savings versus PAYG.
Enterprise: Custom pricing — contact sales for volume, self-hosting, and support options.
Notes & billing mechanics (official):
- Pay As You Go: "No minimums. No expiration. No credit card required." (includes initial $200 credit upon signup).
- Growth: prepaid annual credits redeemed against usage; refunds for unused credits allowed within 30 days per Deepgram billing policy.
- Multichannel billing multiplies single-channel cost by number of channels. Rates shown opt in to Deepgram's Model Improvement Program where noted.
(Extracted only from Deepgram's official pricing pages and documentation.)
Seller details
Deepgram, Inc.
San Francisco, CA, USA
2015
Private
https://deepgram.com
https://x.com/deepgramai
https://www.linkedin.com/company/deepgram/
