TTS & Voice Synthesis
Comparison of voice synthesis solutions for conversational pipelines (2025–2026). Benchmarks, strategic stakes, and decision questions.
Strategic framing: hear TTS in context
What the numbers say
TTFA measures audible start; ELO is a preference signal, not a guarantee for your script. Treat both as signals, then listen to short turns, pauses, numbers and sensitive lines.
What shapes the experience
Cloning, languages and especially performance direction determine how the voice speaks. Distinguish explicit control from simple instructions: test whether intent, pace and pauses hold from one turn to the next.
Stay alert to
Price per hour is an entry point, not a production budget: include volume, streams, caching and regeneration. Also check cloning consent and rights, data residency, and fallback if a voice or API changes.
Reading method: set the required control and voice level first, then compare latency and quality on the same lines, and finally test cost, rights and fallback across a complete session.
This section compares cloud streaming TTS APIs (2025–2026). They speed up integration, with vendor dependency and sovereignty constraints to assess for each deployment context.
Click a header to sort
| Solution | TTFA | ELO | Cloning | Performance direction | Multilingual | Price/hr · entry | Detail |
|---|---|---|---|---|---|---|---|
Eleven v4 / v4 TurboUpd Phase 1 MVP — Référence qualité | 150 ms | 1329 | ✓ | 7/10 · Instruction-based direction | ✓ 90 | $2.40/h | View → |
Cartesia Sonic 3.6Upd Phase 1 MVP — Latence critique | 90 ms | 1285 | ✓ | 10/10 · Explicit API control | ✓ 44 | $2.25/h | View → |
Gemini 3.8 Flash TTSNew Direction de jeu vocale — à tester en français | unpublished | 1275 | ✓ | 7/10 · Instruction-based direction | ✓ 130 | API price pending | View → |
Inworld Realtime TTS-2 + FlashUpd Phase 1 MVP — Qualité + Pipeline conversationnel complet + 100+ langues | 100 ms | 1262 | ✓ | 7/10 · Instruction-based direction | ✓ 200 | $1.35/h | View → |
Gemini 3.8 Flash-Lite TTSNew Cascade temps réel — à chiffrer | unpublished | 1248 | ✓ | 7/10 · Instruction-based direction | ✓ 101 | API price pending | View → |
StepAudio 2.5 TTS Benchmarking qualité — non recommandé pour production EU | 200 ms | 1214 | ✓ | 7/10 · Instruction-based direction | ✓ 100 | $4.59/h | View → |
HeyGen Voice 1New Candidat TTS — Où est Ava ?, sans remplacer Gradium | unpublished | 1196 | ✓ | 7/10 · Instruction-based direction | ✓ 39 | $1.80/h | View → |
Fish Audio S2.1 ProUpd Phase 1 MVP — Coût/Souveraineté | 200 ms | 1190 | ✓ | 7/10 · Instruction-based direction | ✓ 83 | $0.81/h | View → |
Gradium TTSNew Phase B R&D — Lip-sync + Infrastructure Temps Réel | 216 ms | 1174 | ✓ | 3/10 · Limited control | ✓ 5 | $2.60/h | View → |
MiniMax Speech 2.8New Prototypage économique — non recommandé pour production EU | 200 ms | 1173 | ✓ | 10/10 · Explicit API control | ✓ 17 | $3.24/h | View → |
OpenAI Realtime (gpt-realtime-2.1) + GPT-Live-1Upd Dialogue live — délégation GPT-5.5 | 300 ms | 1079 | ✗ | 7/10 · Instruction-based direction | ✓ 70 | $4.61/h | View → |
Deepgram Flux TTSNew Phase B R&D — Test conversationnel temps réel | 80 ms | 1053 | ✗ | 10/10 · Explicit API control | ✗ | $2.70/h | View → |
Hume AI Octave 2Upd Phase 1 MVP — Expressivité émotionnelle | 100 ms | 1051 | ✓ | 10/10 · Explicit API control | ✓ 11 | API price pending | View → |
Smallest.ai Lightning V3.1 Pipeline complète — Phase B Hydra S2S | 100 ms | 1024 | ✓ | 3/10 · Limited control | ✓ 15 | $0.94/h | View → |
Gemini 3.8 LiveNew Dialogue live — hors cascade TTS | unpublished | — | ✗ | 3/10 · Limited control | ✓ 97 | API price pending | View → |
Gemini 3.8 Live Extended ThinkingNew Dialogue live — raisonnement parallèle | unpublished | — | ✗ | 3/10 · Limited control | ✓ 97 | API price pending | View → |
Fish Audio Drama 3New Où est Ava ? — test de direction, hors antenne | unpublished | — | ✓ | 7/10 · Instruction-based direction | ✓ uncounted | API price pending | View → |
Deepgram Aura 2Upd Phase 1 MVP — Stack ASR+TTS intégré | 80 ms | — | ✗ | 3/10 · Limited control | ✗ | $1.62/h | View → |
xAI Grok TTSNew Phase B R&D — Évaluation pipeline S2S | 285 ms | — | ✓ | 7/10 · Instruction-based direction | ✓ 20 | $0.81/h | View → |
Compare by solution
Eleven v4 / v4 Turbo
Eleven v4 (28 sept. 2026) — API $0.08/1k caractères, Turbo $0.04. Arena publique 17 oct. : v4 ELO 1329 (#2), Turbo 1330 (#1).
Instruction-based direction
Cartesia Sonic 3.6
Sonic 3.6 (27 août 2026) — 44 langues, plafond éditeur sous 90 ms. Arena publique 17 oct. : ELO 1285, rang 4.
Explicit API control
Gemini 3.8 Flash TTS
Nouveau — design de voix et acting ligne à ligne, 130 langues. Pas de TTFA publié.
Instruction-based direction
Inworld Realtime TTS-2 + Flash
TTS conversationnel disponible : direction libre, contexte audio, non-verbaux et variante Flash à faible délai serveur
Instruction-based direction
Gemini 3.8 Flash-Lite TTS
Nouveau — variante débit et agents en cascade, 101 langues. Pas de TTFA publié.
Instruction-based direction
StepAudio 2.5 TTS
Contextual TTS — ELO AA V2 Free 1210 (5 octobre ; rang non publié), dual-level context control, zero-shot voice cloning in 3 sec
Instruction-based direction
HeyGen Voice 1
Nouveau — TTS maison HeyGen, octobre 2026. Arena Controlled Voice, pas l’arena vocale principale.
Instruction-based direction
Fish Audio S2.1 Pro
Pay-as-you-go voice cloning — 70% cheaper than ElevenLabs
Instruction-based direction
Gradium TTS
TTS streaming français avec précision de prononciation structurée et résidence UE activable
Limited control
MiniMax Speech 2.8
Sound tags natifs — Speech 2.8 HD ELO 1173 (rang 22). Tarif d’entrée Turbo $60/1M, HD arena $100/1M.
Explicit API control
OpenAI Realtime (gpt-realtime-2.1) + GPT-Live-1
GPT-Live-1 parle, GPT-5.5 raisonne derrière. API au 10 sept. à 0,05 $/min pour la voix, plus le modèle. Scores The Batch contre AVM seulement.
Instruction-based direction
Deepgram Flux TTS
New — conversation-native TTS: 80ms first audio (vendor), cross-turn context, WebSocket + REST
Explicit API control
Hume AI Octave 2
Emotional TTS / speech-to-speech — API retirement announced for 13 Nov. 2026
Explicit API control
Smallest.ai Lightning V3.1
Full voice pipeline (TTS + STT + LLM + S2S) — sub-100ms, 15+ languages
Limited control
Gemini 3.8 Live
Dialogue live (15 sept. 2026) — 97 langues, outils en arrière-plan. Pas un TTS de cascade. Pas de milliseconde publiée.
Limited control
Gemini 3.8 Live Extended Thinking
Dialogue live à raisonnement parallèle (15 sept. 2026). Les indices sont des chiffres Google. Pas de milliseconde publiée.
Limited control
Fish Audio Drama 3
Preview — direct the performance in the script. No published price or time to first audio.
Instruction-based direction
Deepgram Aura 2
Aura 2 baseline — Coval: 327ms median TTFA, 5.1% WER (12 Aug 2026)
Limited control
xAI Grok TTS
TTS naturel et expressif — #3 Humanness Index Vapi (93/100), 460ms TTFA, 5 voix, 20 langues, $15/1M chars
Instruction-based direction
Global architecture: Voice-to-Voice & real time
GPT-Live-1 and Gemini 3.8 Live are dialogue records in this table, not cascade TTS engines. Their architecture comparison — full duplex, delegation, vendor indexes — is on the Voice-to-Voice page. No unpublished millisecond figure is turned into a TTFA.
Explore Voice-to-Voice →