Speech and Voice AI
Frontier, coverage by voice-agent frameworks and the model hub over stt, tts and realtime speech. The Kin Score Market layer reads each provider in its market: who it integrates with, how deep those integrations go, and where the market publishes its own structure. Methodology.
Where this market publishes its structure
- Pipecat services (src/pipecat/services, stt/tts/realtime files) machine-readable
- LiveKit Agents plugins (livekit-plugins/*, stt/tts/realtime files) machine-readable
- Hugging Face /api/models pipeline_tag text-to-speech + automatic-speech-recognition machine-readable, 688 of 700 orgs are research labs, individuals or re-quantizers: counted, never linked
- LiteLLM model_prices json (audio_speech, audio_transcription) machine-readable, router count only; edges owned by sweep-model-routing.py
- Vapi / Retell provider lists not machine-readable: docs pages, not machine-readable
Own = declares at least one integration itself; covered = visible only because an aggregator, standards body, regulator or partner contract names it; distribution-only = listed only on cloud marketplaces it sells through, a channel reported beside, never counted as its own declared integrations; undeclared = nothing we can read, so no number; declares_compatibility = providers with a `compatible` shape vote, report only (kin, 2026-10-13): never counted as own, never in depth or position.
Taxonomy how this market maps onto apis.io
Derived 2026-10-15 from the 49 providers in this market we could resolve in the catalog. The percentage is the share of those providers carrying each one; it describes the market, not the tag or industry. Tag groups have no page of their own yet, so they are listed unlinked.
| Tags | Real-Time 45%, Voice 39%, Text-to-Speech 35%, LLM 29%, Voice AI 27%, Speech-to-Text 27%, Audio 20%, Inference 20%, Transcription 20% |
|---|---|
| Tag groups | Real-Time 47%, Text-to-Speech 43%, Messaging 39%, Conversational AI 39%, LLM 37%, Transcription 33%, Audio 22%, MCP 20% |
| Industries | Artificial Intelligence 98%, Telecommunications & Connectivity 41%, Music, Audio & Creator Economy 22% |
| Areas | Machine Learning 33%, Agents 10% |
| Capabilities | Information & Data Management 24% |
share of resolved cohort carrying it; tags >=0.20 & >=3 members (max 12, junk dropped), tag_groups >=0.20 (cooccurrence, non-cross-cutting; max 8), industries >=0.20 (max 6), areas >=0.10 (max 6; areas are keyword-matched and sparse), capabilities L1 >=0.20 (max 8); derive-market-taxonomy.py
Leaders by position (provisional)
Provisional: grouped by position, depth compared within a position only. Distribution (cloud marketplaces) and iPaaS bridges are counted beside position, never inside it. Unknown-builder listings are left out of position shares and counted here.
covered 48 · reached mainly through aggregators
| Provider | Depth | Integrations | In / out | Unknown builder | Distribution | iPaaS bridge | Mix |
|---|---|---|---|---|---|---|---|
| OpenAI | 53.2 | 47 | 0 / 47 | 0 | 0 | 30 | ipaas-bridge 68%, contract 5%, covered 25%, builds-on-platform 2% |
| ElevenLabs | 51.8 | 5 | 0 / 5 | 0 | 0 | 0 | covered 100% |
| MiniMax | 51.8 | 5 | 0 / 5 | 0 | 0 | 0 | covered 100% |
| Mistral AI | 51.8 | 12 | 0 / 12 | 0 | 1 | 0 | covered 83%, builds-on-platform 8%, distribution 8% |
| Amazon Bedrock | 51.7 | 8 | 0 / 8 | 0 | 0 | 0 | covered 100% |
| Amazon Polly | 51.7 | 3 | 0 / 3 | 0 | 0 | 0 | covered 100% |
| Amazon Transcribe | 51.7 | 3 | 0 / 3 | 0 | 0 | 0 | covered 100% |
| fal | 51.7 | 4 | 0 / 4 | 0 | 0 | 0 | covered 100% |
| Fireworks AI | 51.7 | 7 | 0 / 7 | 0 | 0 | 0 | covered 100% |
| Fish Audio | 51.7 | 3 | 0 / 3 | 0 | 0 | 0 | covered 100% |
| Google Gemini | 51.7 | 7 | 0 / 7 | 0 | 0 | 0 | covered 100% |
| Groq | 51.7 | 7 | 0 / 7 | 0 | 0 | 0 | covered 100% |
| Azure OpenAI Service | 51.7 | 7 | 0 / 7 | 0 | 0 | 0 | covered 100% |
| Neuphonic | 51.7 | 3 | 0 / 3 | 0 | 0 | 0 | covered 100% |
| Nvidia | 51.7 | 4 | 0 / 4 | 0 | 0 | 0 | covered 50%, builds-on-platform 50% |
| NVIDIA NIM | 51.7 | 6 | 0 / 6 | 0 | 0 | 0 | covered 100% |
| Qwen | 51.7 | 3 | 0 / 3 | 0 | 0 | 0 | covered 100% |
| Resemble AI | 51.7 | 3 | 0 / 3 | 0 | 0 | 0 | covered 100% |
| Together AI | 51.7 | 7 | 0 / 7 | 0 | 0 | 0 | covered 100% |
| xAI | 51.7 | 7 | 0 / 7 | 0 | 0 | 0 | covered 100% |
| Deepgram | 47.1 | 5 | 0 / 5 | 0 | 1 | 0 | covered 80%, distribution 20% |
| Cartesia | 46.0 | 4 | 0 / 4 | 0 | 1 | 0 | covered 75%, distribution 25% |
| Baseten | 45.2 | 7 | 0 / 7 | 0 | 2 | 0 | covered 71%, distribution 29% |
| Bland AI | 44.1 | 3 | 0 / 3 | 0 | 1 | 0 | covered 67%, distribution 33% |
| Camb.AI | 44.1 | 3 | 0 / 3 | 0 | 1 | 0 | covered 67%, distribution 33% |
| Smallest AI | 44.1 | 3 | 0 / 3 | 0 | 1 | 0 | covered 67%, distribution 33% |
| Speechify | 44.1 | 3 | 0 / 3 | 0 | 1 | 0 | covered 67%, distribution 33% |
| Sarvam | 42.5 | 5 | 0 / 5 | 0 | 2 | 0 | covered 60%, distribution 40% |
| Gladia | 34.5 thin | 2 | 0 / 2 | 0 | 0 | 0 | covered 100% |
| Google Cloud Speech-To-Text | 34.5 thin | 2 | 0 / 2 | 0 | 0 | 0 | covered 100% |
| Google Cloud Text-To-Speech | 34.5 thin | 2 | 0 / 2 | 0 | 0 | 0 | covered 100% |
| Gradium | 34.5 thin | 2 | 0 / 2 | 0 | 0 | 0 | covered 100% |
| Hume AI | 34.5 thin | 2 | 0 / 2 | 0 | 0 | 0 | covered 100% |
| Inworld AI | 34.5 thin | 2 | 0 / 2 | 0 | 0 | 0 | covered 100% |
| Microsoft Cognitive Services | 34.5 thin | 2 | 0 / 2 | 0 | 0 | 0 | covered 100% |
| Rime | 34.5 thin | 2 | 0 / 2 | 0 | 0 | 0 | covered 100% |
| Speechmatics | 34.5 thin | 2 | 0 / 2 | 0 | 0 | 0 | covered 100% |
| Murf | 26.8 thin | 2 | 0 / 2 | 0 | 1 | 0 | covered 50%, distribution 50% |
| Simplismart | 26.8 thin | 2 | 0 / 2 | 0 | 1 | 0 | covered 50%, distribution 50% |
| AssemblyAI | 26.6 | 17 | 0 / 17 | 0 | 1 | 0 | covered 80%, distribution 20% |
| Argmax | 17.2 thin | 1 | 0 / 1 | 0 | 0 | 0 | covered 100% |
| Fixie | 17.2 thin | 1 | 0 / 1 | 0 | 0 | 0 | covered 100% |
| LMNT | 17.2 thin | 1 | 0 / 1 | 0 | 0 | 0 | covered 100% |
| PyannoteAI | 17.2 thin | 1 | 0 / 1 | 0 | 0 | 0 | covered 100% |
| Respeecher | 17.2 thin | 1 | 0 / 1 | 0 | 0 | 0 | covered 100% |
| SLNG | 17.2 thin | 1 | 0 / 1 | 0 | 0 | 0 | covered 100% |
| Suno | 17.2 thin | 1 | 0 / 1 | 0 | 0 | 0 | covered 100% |
| SYSTRAN | 17.2 thin | 1 | 0 / 1 | 0 | 0 | 0 | covered 100% |
core 1 · builds the connectors its customers need
| Provider | Depth | Integrations | In / out | Unknown builder | Distribution | iPaaS bridge | Mix |
|---|---|---|---|---|---|---|---|
| Telnyx | 48.9 | 8 | 0 / 8 | 0 | 0 | 0 | provider-built 38%, coverage 25%, covered 25%, builds-on-platform 12% |
Frontier timeline DRAFT · report only — not weighted
2026-10-15 anchor shape:
- Shape gravity: 2 providers declare compatibility with the shape in their own contracts.
- Mediation: frameworks and sources cover 49 providers; 2 are covered by every one (mistral, openai).
Generated 2026-10-15 from API Evangelist's own files; nothing on this page is a model's judgement. Provisional and unweighted; not part of the Kin Score composite or agent readiness. Read the Market layer methodology. Disagree with an edge? Every edge cites its evidence and can be corrected through the normal correction path.