Deepgram Aura-2
Multilingual real-time text-to-speech with 90+ voices, pronunciation control and streaming WebSocket delivery for voice agents.
About
Aura-2 is Deepgram's production text-to-speech model for conversational AI, built for the voice-agent pipeline rather than audiobook narration. It streams over REST or WebSocket, ships 90 voices across English, Spanish, German, French, Dutch, Italian and Japanese, and gives runtime control over pronunciation (IPA) and speaking speed for English and Spanish. Deepgram documents self-hosted deployments for supported languages, which matters for teams with data-residency requirements.
Pricing is straightforward: $30 per million characters pay-as-you-go at Deepgram's published rates, sitting between the budget Aura-1 and the newer Flux TTS, which Deepgram now positions as its recommended English voice-agent model. Aura-2 has no published quality benchmarks and the weights are closed, so you are buying an API, not a model. Its advantage is the platform around it: STT, TTS and the Voice Agent API in one place, with conversational turn handling that third-party wrappers can build on.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
No published benchmarks found
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (TIER metrics, metrics-v1). Evidence: GA April 2025 (audioXpress industry coverage); $0.030/1K chars pay-as-you-go (Deepgram pricing via Shunya Labs); 90 voices, 7 languages, IPA pronunciation control (Deepgram skills repo); real production traffic on OpenRouter. Strengths: established speech-infrastructure provider, full STT/TTS/agent platform, self-hosted option documented. Docked for: (1) no published quality benchmarks; (2) closed weights, API-only; (3) partly superseded by Deepgram's own Flux TTS for English voice agents. License: proprietary (commercial API).
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 3.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · GA April 2025 (audioXpress industry coverage)
- · Pay-as-you-go $0.030/1K characters = $30/1M chars (Deepgram pricing via Shunya Labs)
- · 90 voices; EN/ES/DE/FR/NL/IT/JA; IPA pronunciation + speed control (Deepgram skills repo)
- · Real production traffic on OpenRouter (top app ~819K tokens)
- · No published quality benchmarks; closed weights; partly superseded by Deepgram's own Flux TTS for English voice agents