{
  "listings": [
    {
      "slug": "amazon-polly",
      "name": "Amazon Polly",
      "company": "Amazon Web Services, Inc.",
      "grade": "BB",
      "fit": "yes",
      "why": "Speech marks return word, sentence, viseme and SSML mark timings as line-delimited JSON on standard, neural and long-form voices (not generative), billed like speech; prosody rate steers duration; acc-35 drives a lip-synced avatar with them.",
      "source": "https://docs.aws.amazon.com/polly/latest/dg/speechmarks.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/amazon-polly"
    },
    {
      "slug": "azure-text-to-speech",
      "name": "Azure AI Speech text-to-speech",
      "company": "Microsoft Corporation",
      "grade": "BB",
      "fit": "yes",
      "why": "The SDK raises WordBoundary events with the audio offset in ticks and the text position, VisemeReceived gives viseme IDs, SVG or blend shapes, and prosody rate steers duration.",
      "source": "https://learn.microsoft.com/en-us/azure/ai-services/speech-service/how-to-speech-synthesis",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/azure-text-to-speech"
    },
    {
      "slug": "fish-audio-tts",
      "name": "Fish Audio TTS API",
      "company": "Hanabi AI Inc.",
      "grade": "B",
      "fit": "yes",
      "why": "POST /v1/tts/stream/with-timestamp and the /v1/tts/live/with-timestamp WebSocket return word-level alignment with the audio, and prosody.speed runs from 0.5 to 2.0.",
      "source": "https://docs.fish.audio/api-reference/openapi.json",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/fish-audio-tts"
    },
    {
      "slug": "inworld-tts",
      "name": "Inworld Realtime TTS-2",
      "company": "Theai, Inc. dba Inworld AI",
      "grade": "C",
      "fit": "yes",
      "why": "timestampType WORD returns word start and end times with phoneme and viseme detail on every language the model supports, in sync or trailing chunks on streams, and speakingRate from 0.5 to 1.5 sets duration.",
      "source": "https://docs.inworld.ai/tts/capabilities/timestamps",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/inworld-tts"
    },
    {
      "slug": "minimax-speech",
      "name": "MiniMax Speech (T2A)",
      "company": "Nanonoble Pte. Ltd.",
      "grade": "C",
      "fit": "yes",
      "why": "subtitle_enable with subtitle_type word or word_streaming returns word-level timestamps in a file or alongside streamed audio, and voice_setting.speed changes duration.",
      "source": "https://platform.minimax.io/docs/api-reference/speech-t2a-http",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/minimax-speech"
    },
    {
      "slug": "cartesia-tts",
      "name": "Cartesia Sonic TTS API + MCP",
      "company": "Cartesia AI, Inc.",
      "grade": "B",
      "fit": "caveat",
      "why": "add_timestamps returns word start and end times on SSE and WebSocket (acc-02 built highlighted audiobooks on them), but speed 0.6 to 1.5 is guidance rather than a strict multiplier, and acc-30 saw timestamps collapse in Japanese.",
      "source": "https://docs.cartesia.ai/examples/tts-sse-with-timestamps.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/cartesia-tts"
    },
    {
      "slug": "elevenlabs-tts",
      "name": "ElevenLabs Text to Speech API + MCP",
      "company": "Eleven Labs Inc.",
      "grade": "BB",
      "fit": "caveat",
      "why": "The with-timestamps routes and the WebSocket return character-level start and end times (alignment and normalized_alignment), so word timings must be assembled from characters; voice_settings.speed changes duration without a stated range.",
      "source": "https://elevenlabs.io/docs/api-reference/text-to-speech/convert-with-timestamps.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/elevenlabs-tts"
    },
    {
      "slug": "google-cloud-tts",
      "name": "Google Cloud Text-to-Speech",
      "company": "Google LLC",
      "grade": "B",
      "fit": "caveat",
      "why": "Timepoints come back only for SSML mark tags (enableTimePointing SSML_MARK), so per-word timing needs a mark before each word; speaking_rate 0.25 to 2.0 steers Chirp 3 HD, but Gemini-TTS has no rate parameter and acc-15 got 9.5 minutes of audio for a 10-minute video.",
      "source": "https://docs.cloud.google.com/text-to-speech/docs/reference/rest/v1beta1/text/synthesize",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/google-cloud-tts"
    },
    {
      "slug": "heygen-voice",
      "name": "HeyGen Voice",
      "company": "HeyGen Technology, Inc.",
      "grade": "BB",
      "fit": "caveat",
      "why": "with_timestamps returns per-character start and end times on the completed, SSE and WebSocket routes at no extra price, but speed is accepted and ignored on public and instant voices, so duration cannot be steered except with a professional clone.",
      "source": "https://developers.heygen.com/docs/voices/heygen-voice-speech",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/heygen-voice"
    },
    {
      "slug": "murf-tts",
      "name": "Murf TTS API + MCP",
      "company": "Murf Inc.",
      "grade": "BB",
      "fit": "caveat",
      "why": "add_word_timestamps on the WebSocket returns word start and end times in their own frames, and rate sets speed, but timestamps are absent on HTTP streaming and Gen2 and do not work in Chinese, Japanese or Korean.",
      "source": "https://murf.ai/api/docs/text-to-speech/web-sockets/advanced-settings",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/murf-tts"
    },
    {
      "slug": "resemble-ai-tts",
      "name": "Resemble AI Text-to-Speech API",
      "company": "Resemble AI, Inc.",
      "grade": "D",
      "fit": "caveat",
      "why": "audio_timestamps returns grapheme and phoneme start and end pairs on synchronous and WebSocket responses and as WAV cue chunks on HTTP streaming, but there are no word spans and no speed or rate parameter to steer duration.",
      "source": "https://docs.resemble.ai/voice-generation/text-to-speech/synchronous",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/resemble-ai-tts"
    },
    {
      "slug": "rime-tts",
      "name": "Rime TTS API + MCP",
      "company": "Rime Labs, Inc.",
      "grade": "C",
      "fit": "caveat",
      "why": "Coda returns word spans in the done event when timestamps.enable is set and timeScaleFactor 0.4 to 2.5 steers duration, but spans come only for de, en, es, fr, it and pt, not above 180 seconds of audio, and /ws3 timings are estimated from text.",
      "source": "https://docs.rime.ai/api-reference/coda/websocket-v1",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/rime-tts"
    },
    {
      "slug": "smallest-lightning-tts",
      "name": "Smallest.ai Lightning TTS",
      "company": "Smallest, Inc.",
      "grade": "B",
      "fit": "caveat",
      "why": "word_timestamps returns per-word start and end on the WebSocket only, and only for six English and Hindi base-queue voices; the flag is silently ignored elsewhere. speed 0.5 to 2.0 is documented as pitch-free.",
      "source": "https://docs.smallest.ai/models/text-to-speech/word-timestamps.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/smallest-lightning-tts"
    },
    {
      "slug": "soniox-tts",
      "name": "Soniox Text-to-Speech",
      "company": "Soniox Inc.",
      "grade": "BB",
      "fit": "caveat",
      "why": "return_timestamps gives character start and end times on the WebSocket only, ignored over REST, and speed runs 0.7 to 1.3 on both routes, so a 10-minute script can be fitted but only within that range.",
      "source": "https://docs.soniox.com/tts/rt/timestamps.mdx",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/soniox-tts"
    },
    {
      "slug": "audio8-tts-realistic-0-6b",
      "name": "Audio8-TTS-Realistic 0.6B",
      "company": "Edge0 (edge0.ai), which names no legal entity or location on its site. The technical report's authors give their affiliation as AutoArk, and commit metadata in Edge0-AI/Audio8_TTS uses an autoark.ai domain",
      "grade": "D",
      "fit": "no",
      "why": "No timestamps or speed parameter in the card or the serving guide.",
      "source": "https://huggingface.co/Edge0/Audio8-TTS-Realistic-Air",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/audio8-tts-realistic-0-6b"
    },
    {
      "slug": "deepgram-tts",
      "name": "Deepgram Text-to-Speech (Aura-2, Flux TTS)",
      "company": "Deepgram, Inc.",
      "grade": "A",
      "fit": "no",
      "why": "No word timestamps or marks on either transport; the feature matrix lists speed, expressivity and interruption only.",
      "source": "https://developers.deepgram.com/docs/flux-tts/feature-overview.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/deepgram-tts"
    },
    {
      "slug": "fish-audio-drama-3",
      "name": "Fish Audio Drama 3",
      "company": "39 AI, Inc. (the licensor named in the Fish Audio Community License). fish.audio's terms of use name Hanabi AI Inc. as the provider of the hosted services",
      "grade": "C",
      "fit": "no",
      "why": "The fish-speech API server and the card's serving table offer no word timestamps and no speed parameter; the agent would have to align its own output.",
      "source": "https://github.com/fishaudio/fish-speech/blob/main/docs/en/inference.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/fish-audio-drama-3"
    },
    {
      "slug": "fish-audio-s2-pro",
      "name": "Fish Audio S2 Pro",
      "company": "39 AI, Inc. (the licensor named in the Fish Audio Research License). fish.audio's terms of use name Hanabi AI Inc. as the provider of the hosted services",
      "grade": "D",
      "fit": "no",
      "why": "No word timestamps and no speed parameter in the API server or client.",
      "source": "https://github.com/fishaudio/fish-speech/blob/main/docs/en/inference.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/fish-audio-s2-pro"
    },
    {
      "slug": "ming-3-0-tts-0-5b",
      "name": "Ming-3.0-tts-0.5B",
      "company": "inclusionAI (the LICENSE's copyright holder; source headers read Copyright (c) Ant Group)",
      "grade": "C",
      "fit": "no",
      "why": "No timestamps in the output, and speed is only the three levels slow, medium and fast.",
      "source": "https://github.com/inclusionAI/Ming-3.0-tts",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/ming-3-0-tts-0-5b"
    },
    {
      "slug": "ming-3-0-tts-tiny",
      "name": "Ming-3.0-tts-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": "No timestamps in the output; speed is three fixed levels.",
      "source": "https://github.com/inclusionAI/Ming-3.0-tts",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/ming-3-0-tts-tiny"
    },
    {
      "slug": "omnivoice",
      "name": "OmniVoice",
      "company": "k2-fsa, the Next-gen Kaldi open-source project. The paper's authors are at Xiaomi Corp.; no company is named as the licensor",
      "grade": "C",
      "fit": "no",
      "why": "No word timestamps; duration fixes the clip length in seconds and speed scales it, which covers the length target but not the alignment the use case needs.",
      "source": "https://github.com/k2-fsa/OmniVoice",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/omnivoice"
    },
    {
      "slug": "qwen3-tts-1-7b-customvoice",
      "name": "Qwen3-TTS 12Hz 1.7B CustomVoice",
      "company": "Alibaba Cloud (the copyright holder in the Qwen licences; RDAP names Alibaba Cloud (Singapore) Private Limited as registrant of qwen.ai, from the first batch's check)",
      "grade": "C",
      "fit": "no",
      "why": "No timestamps in the output and no speed parameter beyond a plain-language instruct.",
      "source": "https://github.com/QwenLM/Qwen3-TTS",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/qwen3-tts-1-7b-customvoice"
    },
    {
      "slug": "vieneu-tts-v3-turbo",
      "name": "VieNeu-TTS v3 Turbo",
      "company": "Phạm Nguyễn Ngọc Bảo, an individual (named on the card as author; Hugging Face account pnnbao-ump, GitHub account pnnbao97). No company found",
      "grade": "C",
      "fit": "no",
      "why": "No timestamps, and the server accepts speed only for compatibility and ignores it.",
      "source": "https://github.com/pnnbao97/VieNeu-TTS/blob/main/docs/streaming.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/vieneu-tts-v3-turbo"
    }
  ],
  "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": [
      "video.compose-with-audio",
      "subtitles.write-srt"
    ],
    "area": "voice",
    "asks": [
      "narrate this video so it ends with the footage",
      "give me word timings so I can highlight as it reads",
      "drive the avatar's mouth from the speech"
    ],
    "before": [
      "inference.chat-completion",
      "video.describe-frames"
    ],
    "blockers": [
      "no timestamps on the endpoint (Resemble, Murf, most open weights) so the agent must run speech-to-text on its own output",
      "timestamps collapsing on some languages (acc-30 in Japanese), which clipped the end of calls",
      "duration not controllable, so the voiceover misses the video length (acc-15)"
    ],
    "capabilities": [
      "speech.tts",
      "speech.word-timestamps",
      "speech.voices"
    ],
    "category": "text-to-speech",
    "categoryName": "Text-to-speech",
    "checked": "2026-11-12",
    "companies": {
      "fit": 5,
      "fitWithLimit": 9,
      "tagged": 40
    },
    "competition": "crowded",
    "fits": [
      "amazon-polly",
      "azure-text-to-speech",
      "fish-audio-tts",
      "inworld-tts",
      "minimax-speech"
    ],
    "fitsWithLimit": [
      "cartesia-tts",
      "elevenlabs-tts",
      "google-cloud-tts",
      "heygen-voice",
      "murf-tts",
      "resemble-ai-tts",
      "rime-tts",
      "smallest-lightning-tts",
      "soniox-tts"
    ],
    "id": "text-to-speech.timed-to-video",
    "listings": {
      "caveat": 9,
      "no": 9,
      "notRead": 0,
      "tagged": 50,
      "unchecked": 0,
      "yes": 5
    },
    "name": "Produce narration timed to video, animation or highlighted text",
    "status": "draft",
    "success": "The response carries word timestamps (or speech marks and visemes), and total duration can be steered with a speed parameter. Measured: timestamps align within 50 ms on a 30-second check, and a 10-minute script lands within 2 percent of the target.",
    "url": "https://www.anchorterminal.com/api/v1/use-cases/text-to-speech.timed-to-video.json",
    "weight": 8,
    "weightBasis": "6 of 46 accounts tagged (acc-01, 02, 15, 30, 35, 42); a steady minority, and the one case where timestamps rather than voice quality chose the tool"
  }
}
