Skip to content
CallMissed

Explore CallMissed

Back to all models
Text to Speechenglishstreamingvoice-agents
Deepgram — deepgram-flux-tts logo

Deepgram Flux TTS

by Deepgram · Released 2026

Deepgram Flux TTS — streaming-first text-to-speech built for voice agents: turn-based synthesis with prosody carried across turns. 36 English voices including three Indian-accented ones. English only, no expressive controls. Available through the managed Voice Agent, billed in the per-minute voice rate.

Deepgram — deepgram-flux-tts logo
Text to Speech

Deepgram Flux TTS

Powered by Deepgram · Streaming turn-based neural TTS

Context Window

N/A

Parameters

Undisclosed

Max Output

N/A

Category

Text to Speech

Overview

Flux TTS is Deepgram's streaming-first text-to-speech model, designed around the way a voice agent actually speaks: in turns, not in one long block of narration. Synthesis is turn-based, and prosody carries across turns, so the agent's intonation and pacing stay continuous over a conversation instead of resetting on every reply.

Eleven English voices are available: Alexis, Bruce, Cole, Drew, Haley, Heather, Jack, Marcus, Priya, Rufus, and Sharon. Priya is Indian-accented English; the rest cover a range of American English personas.

The model is English-only, with multilingual coverage planned for a later release, and it carries no expressive, emotion, or style controls. On CallMissed, Flux TTS is offered only through the managed Voice Agent — select it on a voice session with tts_engine "flux". It is not a standalone text-to-speech (POST /v1/audio/speech) model. Deepgram serves standalone Flux only on its self-hosted /v2/speak endpoint, which is not part of the hosted API. On the managed Voice Agent, Flux speech is billed inside the per-minute voice rate rather than per character.

Pricing

MetricPrice
Price /10K chars₹0.0000

1 credit = ₹1 = $0.01 USD. Transparent per-call pricing — you pay only for what you call, with no seat fees or minimums.

Key Highlights

  • Streaming-first, turn-based synthesis
  • Prosody carried across turns
  • 36 English voices, including Indian-accented English (Priya, Meena, Naveen)
  • Managed Voice Agent only — billed in the per-minute voice rate

Benchmarks

BenchmarkScore
Voices11
LanguagesEnglish
AccessVoice Agent

Technical Details

  • Streaming-first text-to-speech built for voice agents
  • Turn-based synthesis with prosody carried across turns
  • Voices: Alexis, Bruce, Cole, Drew, Haley, Heather, Jack, Marcus, Priya, Rufus, Sharon
  • Priya is Indian-accented English
  • English only — multilingual coverage planned for a later release
  • No expressive, emotion, or style controls
  • Offered only through the managed Voice Agent (tts_engine "flux"); not a /v1/audio/speech model

Strengths

  • Turn-based streaming keeps first-audio latency low for conversation
  • Prosody continuity across turns — the agent does not reset its delivery every reply
  • 36 English voices including three Indian-accented options

Limitations

  • English only — multilingual coverage is planned for a later release
  • No expressive, emotion, or style controls
  • Managed Voice Agent only — not available on the /v1/audio/speech endpoint
  • Paid plans only (not in the free tier)

Use Cases

Voice agent speech outputConversational IVRCustomer service botsReal-time English readouts

API Example

curl https://api.callmissed.com/v1/voice/sessions \
  -H "Authorization: Bearer cm_YOUR_KEY" \
  -d '{"tts_engine": "flux", "voice": "priya"}'

Endpoint: POST /v1/voice/sessions (tts_engine: "flux") · Model ID: deepgram-flux-tts

Try Deepgram Flux TTS now

Get 1000 free API credits on signup. No credit card required.

Start free Book a DemoRead docs