{
  "listings": [
    {
      "slug": "assemblyai-stt",
      "name": "AssemblyAI Speech-to-Text (Universal)",
      "company": "AssemblyAI, Inc.",
      "grade": "B",
      "fit": "yes",
      "why": "Turn messages carry end_of_turn true; Terminate returns in-flight turns then Termination. Pipecat measured 282 to 307 ms median to final. acc-31's stall was on the retired v2 transcriber.",
      "source": "https://www.assemblyai.com/docs/streaming/message-sequence",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/assemblyai-stt"
    },
    {
      "slug": "cartesia-ink-stt",
      "name": "Cartesia Ink",
      "company": "Cartesia AI, Inc.",
      "grade": "B",
      "fit": "yes",
      "why": "Ink 2 emits turn.start, turn.eager_end, turn.resume and turn.end; close flushes remaining audio and returns a done acknowledgment. Pipecat measured 299 ms median, 1,584 ms at p99.",
      "source": "https://docs.cartesia.ai/use-the-api/stt/turns",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/cartesia-ink-stt"
    },
    {
      "slug": "elevenlabs-scribe",
      "name": "ElevenLabs Scribe Speech to Text API",
      "company": "Eleven Labs Inc.",
      "grade": "BB",
      "fit": "yes",
      "why": "committed_transcript is the final per segment; commit_strategy turn_prediction on scribe_v3_live commits when the model predicts the turn ended. Pipecat measured scribe_v2_realtime at 281 ms median.",
      "source": "https://elevenlabs.io/docs/eleven-api/guides/how-to/speech-to-text/realtime/event-reference",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/elevenlabs-scribe"
    },
    {
      "slug": "inworld-stt",
      "name": "Inworld STT",
      "company": "Theai, Inc. dba Inworld AI",
      "grade": "C",
      "fit": "yes",
      "why": "Finals carry isFinal true per turn with endOfTurnConfidenceThreshold and client endTurn; closeStream finalises pending audio. No independent latency row, and headers rather than query carry the key.",
      "source": "https://docs.inworld.ai/api-reference/sttAPI/speechtotext/transcribe-stream-websocket",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/inworld-stt"
    },
    {
      "slug": "soniox-stt",
      "name": "Soniox Speech-to-Text",
      "company": "Soniox Inc.",
      "grade": "B",
      "fit": "yes",
      "why": "enable_endpoint_detection finalises the segment and returns an \u003cend\u003e token; finalize forces a final. Pipecat measured stt-rt-v6 at 266 ms median; streams cap at 300 minutes.",
      "source": "https://docs.soniox.com/stt/rt/endpoint-detection",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/soniox-stt"
    },
    {
      "slug": "speechmatics-stt",
      "name": "Speechmatics Speech-to-Text",
      "company": "Cantab Research Ltd",
      "grade": "BB",
      "fit": "yes",
      "why": "Agent STT emits StartOfTurn and EndOfTurn, with AddSegment finals and EndOfStream then EndOfTranscript on close; Pipecat measured linden-1 at 369 ms median. Temporary keys travel in the WebSocket URL.",
      "source": "https://docs.speechmatics.com/speech-to-text/agent-stt/turn-detection",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/speechmatics-stt"
    },
    {
      "slug": "amazon-transcribe",
      "name": "Amazon Transcribe",
      "company": "Amazon Web Services, Inc.",
      "grade": "BB",
      "fit": "caveat",
      "why": "StartStreamTranscription returns segments with IsPartial false, but no end-of-turn event, and the Pipecat benchmark measured 1,136 ms median time to final, above the 500 ms target.",
      "source": "https://docs.aws.amazon.com/transcribe/latest/APIReference/API_streaming_StartStreamTranscription.html",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/amazon-transcribe"
    },
    {
      "slug": "azure-speech-to-text",
      "name": "Azure AI Speech speech-to-text",
      "company": "Microsoft Corporation",
      "grade": "BB",
      "fit": "caveat",
      "why": "Speech SDK recognized events are per-segment finals with no turn event; MAI-Transcribe-2-Streaming has no server VAD and needs a client commit. Pipecat measured 1,016 ms median to final.",
      "source": "https://learn.microsoft.com/en-us/azure/ai-services/speech-service/how-to-recognize-speech",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/azure-speech-to-text"
    },
    {
      "slug": "deepgram-stt",
      "name": "Deepgram Speech-to-Text (Nova-3, Flux)",
      "company": "Deepgram, Inc.",
      "grade": "BB",
      "fit": "caveat",
      "why": "Flux emits EndOfTurn with eot_threshold and eot_timeout_ms; Nova-3 streaming gives speech_final and CloseStream flushes. Pipecat measured Nova-3 at 247 ms, but acc-28 reports sessions silently stopping with no error.",
      "source": "https://developers.deepgram.com/docs/flux/configuration",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/deepgram-stt"
    },
    {
      "slug": "giggy-stt",
      "name": "Giggy Speech to Text",
      "company": "Giggy",
      "grade": "E",
      "fit": "caveat",
      "why": "The live WebSocket returns cumulative transcription events with is_final and input_complete, but the docs say turn boundaries and VAD are the caller's job; no end-of-turn event and no independent latency row.",
      "source": "https://giggy.ai/docs/speech-api",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/giggy-stt"
    },
    {
      "slug": "gladia-stt",
      "name": "Gladia Speech-to-Text API + MCP",
      "company": "Gladia SAS",
      "grade": "BB",
      "fit": "caveat",
      "why": "Live init has endpointing (silence seconds) and maximum_duration_without_endpointing, and receive_final_transcripts sends finals; no end-of-turn event beyond the final, and Gladia has no Pipecat latency row.",
      "source": "https://docs.gladia.io/api-reference/v2/live/init",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/gladia-stt"
    },
    {
      "slug": "google-speech-to-text",
      "name": "Google Cloud Speech-to-Text",
      "company": "Google LLC",
      "grade": "B",
      "fit": "caveat",
      "why": "StreamingRecognize has SPEECH_ACTIVITY_END voice events and is_final results, but streams stop at 5 minutes and must be reopened; Pipecat measured latest-long at 878 ms and gemini-3.5-transcribe-live at 458 ms.",
      "source": "https://docs.cloud.google.com/speech-to-text/docs/reference/rpc/google.cloud.speech.v2",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/google-speech-to-text"
    },
    {
      "slug": "ibm-watson-stt",
      "name": "IBM Watson Speech to Text",
      "company": "International Business Machines Corporation",
      "grade": "C",
      "fit": "caveat",
      "why": "The WebSocket returns final results per phrase with interim_results and end_of_phrase_silence_time, no explicit end-of-turn event, and the access token travels in the URL query string; no independent latency row.",
      "source": "https://cloud.ibm.com/docs/speech-to-text?topic=speech-to-text-websockets",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/ibm-watson-stt"
    },
    {
      "slug": "mistral-voxtral-transcribe",
      "name": "Mistral Voxtral Transcribe",
      "company": "Mistral AI (RCS Paris 952 418 325)",
      "grade": "B",
      "fit": "caveat",
      "why": "Realtime 2602 streams text deltas and a transcription.done event at the end, with no end-of-turn event; Pipecat measured it at 525 ms median. Realtime 3 (Public Preview) claims end-of-turn detection.",
      "source": "https://docs.mistral.ai/studio/audio/speech_to_text/realtime_transcription",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/mistral-voxtral-transcribe"
    },
    {
      "slug": "nemotron-3-5-asr-streaming-0-6b",
      "name": "Nemotron 3.5 ASR Streaming 0.6B",
      "company": "NVIDIA Corporation",
      "grade": "C",
      "fit": "caveat",
      "why": "Cache-aware streaming at 80 to 1,120 ms chunks; Pipecat measured 236 ms median to final, but the model emits text, not an end-of-turn event, so the caller supplies VAD and endpointing.",
      "source": "https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/nemotron-3-5-asr-streaming-0-6b"
    },
    {
      "slug": "openai-speech-to-text",
      "name": "OpenAI Speech to Text",
      "company": "OpenAI OpCo, LLC",
      "grade": "BB",
      "fit": "caveat",
      "why": "gpt-live-transcribe needs the client to commit each turn (no server_vad), then sends completed; acc-21 saw deltas arrive with the final. Pipecat measured gpt-4o-transcribe at 637 ms median.",
      "source": "https://developers.openai.com/api/docs/guides/realtime-transcription",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/openai-speech-to-text"
    },
    {
      "slug": "qwen3-asr-0-6b",
      "name": "Qwen3-ASR-0.6B",
      "company": "Alibaba Cloud (the registrant of qwen.ai per RDAP, carried from the Qwen3.8-27B check)",
      "grade": "C",
      "fit": "caveat",
      "why": "Streaming runs only through vLLM's realtime path, without timestamps or batching, and emits text rather than an end-of-turn event; no independent latency row.",
      "source": "https://huggingface.co/Qwen/Qwen3-ASR-0.6B",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/qwen3-asr-0-6b"
    },
    {
      "slug": "qwen3-asr-1-7b",
      "name": "Qwen3-ASR-1.7B",
      "company": "Alibaba Cloud (the registrant of qwen.ai per RDAP, carried from the Qwen3.8-27B check)",
      "grade": "B",
      "fit": "caveat",
      "why": "Streaming runs only through vLLM's realtime path, without timestamps or batching, and emits text rather than an end-of-turn event; no independent latency row.",
      "source": "https://huggingface.co/Qwen/Qwen3-ASR-1.7B",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/qwen3-asr-1-7b"
    },
    {
      "slug": "rev-ai-stt",
      "name": "Rev AI Speech-to-Text API",
      "company": "Rev.com, Inc.",
      "grade": "C",
      "fit": "caveat",
      "why": "Streaming returns partial and final hypotheses with no end-of-turn event, the access token travels in the URL query string, and Rev AI has no independent latency row.",
      "source": "https://docs.rev.ai/api/streaming/requests",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/rev-ai-stt"
    },
    {
      "slug": "smallest-pulse-stt",
      "name": "Smallest.ai Pulse STT",
      "company": "Smallest, Inc.",
      "grade": "B",
      "fit": "caveat",
      "why": "Live WebSocket has endpointing with eou_timeout_ms, is_final transcripts, speech_ended events and finalize and close_stream messages; Pipecat measured pulse at 398 ms median but 1,593 ms at p99, and keys may travel in the query string.",
      "source": "https://docs.smallest.ai/asyncapi/models-api.json",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/smallest-pulse-stt"
    },
    {
      "slug": "voxtral-mini-4b-realtime-2602",
      "name": "Voxtral Mini 4B Realtime 2602",
      "company": "Mistral AI (RCS Paris 952 418 325)",
      "grade": "B",
      "fit": "caveat",
      "why": "vLLM serves it on /v1/realtime with a 480 ms delay, but it emits text only, no end-of-turn event; Pipecat measured the hosted twin at 525 ms median and 1,913 ms at p99.",
      "source": "https://huggingface.co/mistralai/Voxtral-Mini-4B-Realtime-2602",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/voxtral-mini-4b-realtime-2602"
    },
    {
      "slug": "cohere-transcribe-03-2026",
      "name": "Cohere Transcribe 03-2026",
      "company": "Cohere Inc.",
      "grade": "C",
      "fit": "no",
      "why": "Batch model with no streaming runtime or endpoint; Cohere's docs list audio-in text-out only.",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/cohere-transcribe-03-2026"
    },
    {
      "slug": "groq-speech-to-text",
      "name": "Groq Speech-to-Text",
      "company": "Groq LLC",
      "grade": "BB",
      "fit": "no",
      "why": "No streaming or realtime endpoint; files or URLs only.",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/groq-speech-to-text"
    },
    {
      "slug": "ming-3-0-asr-tiny",
      "name": "Ming-3.0-asr-tiny",
      "company": "inclusionAI, Ant Group (named in the site footer of inclusion-ai.org; the LICENSE copyright line reads inclusionAI)",
      "grade": "C",
      "fit": "no",
      "why": "Batch model; no streaming runtime, only SGLang-Omni server-sent events of a finished transcript.",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/ming-3-0-asr-tiny"
    },
    {
      "slug": "ming-3-0-audio-flash",
      "name": "Ming-3.0-audio-flash",
      "company": "inclusionAI, Ant Group (named in the inclusion-ai.org footer; no registered company named)",
      "grade": "E",
      "fit": "no",
      "why": "Offline audio-language model; no streaming transcription path is documented.",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/ming-3-0-audio-flash"
    },
    {
      "slug": "ming-3-0-audio-tiny",
      "name": "Ming-3.0-audio-tiny",
      "company": "inclusionAI (Ant Group)",
      "grade": "D",
      "fit": "no",
      "why": "Offline audio-language model; no streaming transcription documented.",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/ming-3-0-audio-tiny"
    },
    {
      "slug": "minimax-stt",
      "name": "MiniMax Speech-to-Text",
      "company": "Nanonoble Pte. Ltd.",
      "grade": "D",
      "fit": "no",
      "why": "Files only; stream=true streams text output of an uploaded file, with no live audio input.",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/minimax-stt"
    },
    {
      "slug": "parakeet-tdt-0-6b-v3",
      "name": "Parakeet TDT 0.6B v3",
      "company": "NVIDIA Corporation",
      "grade": "B",
      "fit": "no",
      "why": "An offline TDT model with no streaming runtime or end-of-turn signal; the speech.streaming tag is not supported by the card.",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/parakeet-tdt-0-6b-v3"
    },
    {
      "slug": "underdog-bark-3",
      "name": "Underdog Bark 3.0",
      "company": "Conway Research",
      "grade": "E",
      "fit": "no",
      "why": "Offline model run with Conway's loader; no streaming path.",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/underdog-bark-3"
    }
  ],
  "marks": "yes: the docs show it does the job as the success test says. caveat: it does the job with a limit an agent should know. no: it doesn't do this job. unchecked: we couldn't read it.",
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "attributionRequired": true,
    "credit": "CC BY 4.0, credit Anchor Terminal and link the page or API address you took it from (https://www.anchorterminal.com/terms/#data)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-11-16",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.8",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run",
    "terms": "https://www.anchorterminal.com/terms/#data"
  },
  "useCase": {
    "accounts": 6,
    "after": [
      "llm.reply",
      "text-to-speech.speak"
    ],
    "area": "voice",
    "asks": [
      "transcribe what the caller is saying as they speak",
      "give me the final text when they stop talking",
      "wire speech-to-text into my phone agent"
    ],
    "before": [
      "telephony.answer-call",
      "voice.capture-microphone"
    ],
    "blockers": [
      "no streaming endpoint (Groq, Cohere Transcribe, Ming, Parakeet TDT without a streaming runtime)",
      "stream drops after 5 minutes without reconnect (Google 5-minute stream limit)",
      "session stops returning text with no error, as in acc-28 and acc-31",
      "credentials in the WebSocket query string where logs can keep them"
    ],
    "capabilities": [
      "speech.stt",
      "speech.streaming",
      "speech.turn-detection"
    ],
    "category": "speech-to-text",
    "categoryName": "Speech-to-text",
    "checked": "2026-11-12",
    "companies": {
      "fit": 6,
      "fitWithLimit": 13,
      "tagged": 50
    },
    "competition": "crowded",
    "fits": [
      "assemblyai-stt",
      "cartesia-ink-stt",
      "elevenlabs-scribe",
      "inworld-stt",
      "soniox-stt",
      "speechmatics-stt"
    ],
    "fitsWithLimit": [
      "amazon-transcribe",
      "azure-speech-to-text",
      "deepgram-stt",
      "giggy-stt",
      "gladia-stt",
      "google-speech-to-text",
      "ibm-watson-stt",
      "mistral-voxtral-transcribe",
      "nemotron-3-5-asr-streaming-0-6b",
      "openai-speech-to-text",
      "qwen3-asr-0-6b",
      "qwen3-asr-1-7b",
      "rev-ai-stt",
      "smallest-pulse-stt",
      "voxtral-mini-4b-realtime-2602"
    ],
    "id": "speech-to-text.voice-agent-live",
    "listings": {
      "caveat": 15,
      "no": 8,
      "notRead": 3,
      "tagged": 69,
      "unchecked": 0,
      "yes": 6
    },
    "name": "Transcribe a live caller for a voice agent, with a clear end of turn",
    "status": "draft",
    "success": "A final transcript event for each turn, with an explicit end-of-turn or finalisation signal, in a session that can last a 10-minute call. Measured: the final event arrives within 500 ms of end of speech at the median, and no session silently stops returning text over a 10-minute call.",
    "url": "https://www.anchorterminal.com/api/v1/use-cases/speech-to-text.voice-agent-live.json",
    "weight": 18,
    "weightBasis": "6 of 32 accounts (acc-04, acc-21, acc-28, acc-29, acc-31, acc-11) plus two independent benchmarks built for this job alone; the use case vendors lead with"
  }
}
