Convert text to natural speech using Deepgram Aura-2 TTS, paid per call via MPP/Tempo.
What it does
This endpoint provides text-to-speech synthesis powered by Deepgram's Aura-2 models, accessed through the Locus MPP (Micropayment Protocol) gateway. You send a JSON body with the text you want spoken and receive audio back in your chosen format (MP3, WAV, FLAC, Opus, AAC, and others). Multiple voice models are available in the `aura-2-{voice}-{lang}` naming scheme, including voices like thalia, apollo, and orion for English.
Payment is handled per-call via the Tempo method on pathUSD. Each TTS request costs 23,000 base units of pathUSD (6 decimals), which works out to $0.023 per call. The recommended maximum input is approximately 2,000 characters per request. You can control the output encoding, container format, and sample rate to match your application's requirements.
This endpoint is part of a broader Deepgram service suite on Locus MPP that also includes audio transcription (Nova-3 STT), text analysis (sentiment, topics, intents, summarization), and a model listing endpoint. The TTS endpoint specifically accepts POST requests with a JSON body. Note that the probe returned 404 on HEAD/GET, which is expected since this is a POST-only endpoint — it is not evidence the endpoint is down.
Capabilities
Use cases
- —Generating spoken audio from text for accessibility features
- —Adding voice narration to chatbot or virtual assistant responses
- —Producing audio previews of written content
- —Creating voice prompts for IVR or telephony systems
- —Building read-aloud features in apps or browser extensions
Fit
Best for
- —Agents needing on-demand speech synthesis without API key management
- —Per-call pay-as-you-go TTS with no subscription commitment
- —Applications requiring multiple output audio formats and sample rates
Not for
- —Real-time streaming TTS over WebSocket connections (this is a request/response endpoint)
- —Very long documents exceeding ~2000 characters per call without chunking
- —Free or zero-cost speech synthesis needs
Quick start
curl -X POST https://deepgram.mpp.paywithlocus.com/deepgram/speak \
-H "Content-Type: application/json" \
-d '{"text": "Hello, this is a test of Deepgram text to speech.", "model": "aura-2-thalia-en", "encoding": "mp3"}' \
--output speech.mp3Example
Request
{
"text": "Hello, this is a test of Deepgram text to speech.",
"model": "aura-2-thalia-en",
"encoding": "mp3",
"container": "mp3",
"sample_rate": "24000"
}Endpoint
Quality
Full OpenAPI schema with request body details and pricing is available. However, the probe returned 404 because HEAD/GET were used on a POST-only endpoint, so liveness via 402 challenge was not directly confirmed. No response schema or example response is documented. Crawled pages returned no useful documentation.
Warnings
- —Probe returned 404 on HEAD and GET; endpoint is POST-only so 402 challenge was not captured — liveness not directly confirmed
- —No response schema documented; audio binary output is inferred from context
- —Currency address 0x20c000000000000000000000b9537d11c60e8b50 assumed to be pathUSD with 6 decimals based on Tempo convention; if different, price calculation ($0.023) may be incorrect
Citations
- —TTS endpoint costs 23,000 base units via Tempo methodhttps://deepgram.mpp.paywithlocus.com
- —Aura-2 voice models available in format aura-2-{voice}-{lang}https://deepgram.mpp.paywithlocus.com
- —Supported encodings include mp3, linear16, flac, mulaw, alaw, opus, aachttps://deepgram.mpp.paywithlocus.com
- —Max ~2000 chars recommended per TTS callhttps://deepgram.mpp.paywithlocus.com
- —API reference available at Deepgram developer docshttps://developers.deepgram.com/reference/deepgram-api-overview