{ "data": { "a": { "slug": "fish-audio-drama-3", "name": "Fish Audio Drama 3", "vendor": "Fish Audio", "vendorUrl": "https://fish.audio", "kind": "weights", "category": "text-to-speech", "summary": "Fish Audio Drama 3 is Fish Audio's open-weight text-to-speech model, a 35B mixture of experts with 3B active built on Qwen3.5-35B-A3B, that clones voices, takes plain-language direction and voices multi-speaker scenes, under the Fish Audio Community License.", "url": "https://www.anchorterminal.com/tools/fish-audio-drama-3", "markdownUrl": "https://www.anchorterminal.com/tools/fish-audio-drama-3.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/fish-audio-drama-3.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/fish-audio-drama-3.json", "repo": "https://huggingface.co/fishaudio/drama-3", "license": "Fish Audio Community License 1.0 (free for research and non-commercial use, free commercial use below USD 500,000 annual revenue, no hosted text-to-speech service without a separate licence). The card's metadata still names fish-audio-research-license", "transports": [], "packages": [ { "registry": "huggingface", "name": "fishaudio/drama-3" }, { "registry": "huggingface", "name": "fishaudio/drama-3-int4" } ], "auth": "none", "authNotes": "The weights download from Hugging Face with no account, token or terms to accept. Neither fishaudio/drama-3 nor fishaudio/drama-3-int4 is gated (checked 2026-10-23). The licence says using or accessing the files binds you to it. The fish-speech code is public on GitHub.", "pricing": "free", "pricingNotes": "Free to download. Under the Fish Audio Community License 1.0 (13 October 2026), research and non-commercial use is free, commercial use is free while you and your affiliates have under USD 500,000 annual revenue, and offering it as a hosted text-to-speech, voice or cloning service needs a separate written licence at any size. Running it costs the hardware, an NVIDIA GPU with at least 48 GB for the FP8 build per Fish Audio's install guide. Fish Audio's own API sells `drama-3` at $50 per million UTF-8 bytes, $10 until 12 November 2026 (checked 2026-10-23).", "priceSummary": "Free", "where": "local", "x402": { "level": "no", "evidence": "Weights you download and run yourself, so there is nothing to pay per call and no payment protocol (checked 2026-10-23).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": null, "pypiWeekly": null, "asOf": "2026-10-23" }, "docsUrl": "https://speech.fish.audio/inference/", "capabilities": [ "speech.tts", "speech.languages", "speech.streaming", "voice.clone", "model.open-weights" ], "tags": [ "open-weights", "model", "moe", "text-to-speech", "voice-cloning", "multi-speaker", "speech-editing", "inline-tags", "multilingual", "revenue-threshold", "fp8", "int4", "safetensors", "self-hosted", "new" ], "lastRelease": "2026-10-17", "graded": true, "anchor": { "graded": true, "score": 55.6, "grade": "C", "agentReady": false, "rank": 0, "ranked": false, "notRankedWhy": "An open-weight model, graded by the open-weight checklist and ranked among open-weight models, not against tools", "rankOf": 1201, "categoryRank": 15, "methodology": "0.6", "run": "2026-10-01", "scores": { "capability": 67, "interface": 70, "reach": 60, "risk": 25, "upkeep": 51 }, "pending": [ "performance", "tasks" ], "assessment": { "confidence": "medium", "date": "2026-10-23" }, "negative": 0, "verdict": "A local voice cousin for directed speech, multi-character scenes, voice cloning and word-level edits in about 80 languages. It needs an NVIDIA GPU, about 29 GB for the int4 build or at least 48 GB for FP8, through SGLang-Omni or fish-speech. The main risk is the licence, revocable, capped at USD 500,000 revenue for commercial use and barring hosted services.", "bestFor": "Agents with an NVIDIA GPU of 32 GB or more producing directed multi-character audio, dubbing or narration, for organisations under USD 500,000 revenue or non-commercial use", "strengths": [ "Up to 16 voices per request, cloning from several references, revision by note and word-level editing by infilling, all in one model", "Independently run on the Open TTS Leaderboard, with an English WER of 1.19% without cloning and 2.66% with it", "An int4 build in Fish Audio's organisation that roughly halves the files to 24.0 GB", "Every prompt format and control is written out in the technical report, and SGLang-Omni serves it behind an OpenAI-style speech endpoint with streaming", "Hashed files in Fish Audio's own organisation, ungated" ], "weaknesses": [ "Commercial use is free only below USD 500,000 annual revenue, hosted services always need a separate licence, and the grant is revocable", "The card's licence metadata still names the Research License with a broken link, after a release-day licence switch", "NVIDIA only, and not in any released version of Transformers, vLLM, SGLang, llama.cpp, Ollama or LM Studio", "The codec ships as a PyTorch pickle, and there are no safety results or security contact", "No reply from Fish Audio to any of its 6 Hugging Face discussions" ], "agentNotes": [ "Install SGLang-Omni from source with `uv pip install --prerelease=allow -e \".[fishaudio]\"`, then `sgl-omni serve --model-path fishaudio/drama-3-int4 --port 8000` (or fishaudio/drama-3 for FP8 on 48 GB or more)", "POST JSON to /v1/audio/speech with `input`, optional `references` (audio_path and its transcript) and `previous_takes`; set `response_format` and `stream` for the output", "Start each line with `\u003c|speaker:N|\u003e` (N from 0 to 15) for several voices, and separate lines with `\u003c|audio_separator|\u003e`", "Direct with point tags like `[sighs]`, scoped tags like `\u003cwhisper\u003e...\u003c/whisper\u003e`, a bracketed style prefix, or `Direction:` and `Line:` for a director's note", "The int4 build needs compute capability 8.0 or newer; only the triton attention backend is accepted", "Keep each reference clip under 60 seconds and the prompt under 8,192 rows, about 21.5 rows per second of reference audio", "Check the licence before commercial use or exposing it to other users, and only clone voices with consent", "Treat text from untrusted sources as a script that can carry directions and speaker tags" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 4, "avgRating": 2.8, "history": [ { "basis": "public evidence", "confidence": "medium", "grade": "C", "methodology": "0.6", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 55.6 } ], "editorialScores": { "capability": 67, "interface": 70, "reach": 60, "risk": 25, "upkeep": 51 }, "provenanceScore": 63, "checked": { "categories": { "capability": "2026-10-23", "interface": "2026-10-23", "reach": "2026-10-23", "risk": "2026-10-23", "upkeep": "2026-10-23" }, "last": "2026-10-23", "oldest": "2026-10-23", "stale": false, "staleAfterDays": 90 } }, "connect": { "install": "git clone https://github.com/sgl-project/sglang-omni.git \u0026\u0026 cd sglang-omni\nuv venv -p 3.12 .venv \u0026\u0026 . .venv/bin/activate\nuv pip install --prerelease=allow -e \".[fishaudio]\"\nhf download fishaudio/drama-3-int4 # or fishaudio/drama-3 for the FP8 build\nsgl-omni serve --model-path fishaudio/drama-3-int4 --port 8000", "http": "curl -X POST http://localhost:8000/v1/audio/speech \\\n -H \"Content-Type: application/json\" \\\n -d '{\"input\": \"\u003c|speaker:0|\u003e [sighs] Fine. If we miss it, it is on you.\", \"references\": [{\"audio_path\": \"/path/to/reference.wav\", \"text\": \"Transcript of the reference clip.\"}], \"response_format\": \"wav\"}' \\\n --output out.wav" }, "letme": { "capability": "https://letme.dev/speech.tts", "tool": "https://letme.dev/fish-audio-drama-3" }, "sameCompany": [ "fish-audio-s2-pro", "fish-audio-tts", "fish-audio-voice-cloning" ], "weights": { "trainedBy": "Fish Audio", "builtOn": "Qwen/Qwen3.5-35B-A3B (named in the technical report as the backbone it's initialised from, not on the card; Apache 2.0)", "params": "35.9B (35B backbone with about 3B active per token, per the card; 35,899,643,008 by the Hub's count)", "context": 16384, "goodFor": "Directed, expressive speech for multi-character scenes, long-form narration and dubbing, with voice cloning, revision by note and word-level editing, on an NVIDIA GPU.", "runtimes": [], "builds": [ { "name": "drama-3-int4", "url": "https://huggingface.co/fishaudio/drama-3-int4", "who": "publisher", "precision": "int4", "sizeGB": 24, "memoryGB": 32, "format": "safetensors", "note": "W4A16 GPTQ-format experts and attention projections, round-to-nearest with no calibration, quantised by AK (akhaliq) and duplicated into the fishaudio organisation on 14 October 2026; its card calls it unofficial. 22.1 GB of safetensors plus codec.pth, excluding the evaluation audio. Needs compute capability 8.0 or newer" }, { "name": "drama-3", "url": "https://huggingface.co/fishaudio/drama-3", "who": "publisher", "precision": "fp8", "sizeGB": 41.5, "memoryGB": 64, "format": "safetensors", "note": "The release itself, MoE experts in block-scaled FP8 with the rest in BF16, plus the 1.87 GB codec.pth pickle. Fish Audio's install guide asks for at least 48 GB on an NVIDIA GPU" } ], "run": "git clone https://github.com/sgl-project/sglang-omni.git \u0026\u0026 cd sglang-omni \u0026\u0026 uv pip install --prerelease=allow -e \".[fishaudio]\" \u0026\u0026 hf download fishaudio/drama-3-int4 \u0026\u0026 sgl-omni serve --model-path fishaudio/drama-3-int4 --port 8000", "runBuild": "drama-3-int4", "checked": "2026-10-23", "minMemoryGB": 32 }, "family": { "lab": "Fish Audio", "family": "Drama", "url": "https://www.anchorterminal.com/open-weights/#fishaudio", "olderCount": 8 }, "alsoIn": [ "voice-cloning" ], "area": "voice", "provenance": { "legalEntity": "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", "domain": "fish.audio", "domainRegistered": "2023-12-11", "endpointOnVendorDomain": null, "terms": "https://huggingface.co/fishaudio/drama-3/blob/main/LICENSE", "privacy": "", "statusPage": "", "changelog": "https://huggingface.co/fishaudio/drama-3/commits/main", "securityTxt": "none", "checked": "2026-10-23", "notes": [ "The files sit in the fishaudio organisation on Hugging Face, which github.com/fishaudio/fish-speech, the technical report and Fish Audio's launch post link to.", "Nothing is hosted for this listing, so the licence stands in for terms of service and no privacy policy is scored.", "The Hugging Face history is nine commits from 13 to 17 October 2026 (the upload, the licence file and its rename, card edits, the report link and vLLM-Omni instructions). The weights haven't changed since the first commit. fishaudio/drama-3-int4 has two commits on 14 October 2026.", "fish.audio/.well-known/security.txt answered 404 on 23 October 2026 and fishaudio/fish-speech has no SECURITY.md. The domain registration date is carried from the fish-audio-tts check of 17 October 2026." ], "score": 63 }, "pageJsonUrl": "https://www.anchorterminal.com/tools/fish-audio-drama-3.json", "live": { "slug": "fish-audio-drama-3", "securityTxt": { "url": "https://fish.audio/.well-known/security.txt", "state": "none", "checkedAt": "2026-10-23T15:24:36.233836361Z" }, "updatedAt": "2026-10-23T15:24:36.233836361Z" } }, "answer": "OmniVoice scores 56.9 (C) on agent readiness against Fish Audio Drama 3's 55.6 (C), and leads in 2 of 5 scored categories. Fish Audio Drama 3 leads on reach and agent interface.", "b": { "slug": "omnivoice", "name": "OmniVoice", "vendor": "k2-fsa (Next-gen Kaldi)", "vendorUrl": "https://k2-fsa.org", "kind": "weights", "category": "text-to-speech", "summary": "OmniVoice is k2-fsa's open-weight zero-shot text-to-speech model for over 600 languages, initialised from Qwen3-0.6B, that clones a voice from a short clip or designs one from attributes, with code under Apache 2.0 and weights under CC-BY-NC by the card.", "url": "https://www.anchorterminal.com/tools/omnivoice", "markdownUrl": "https://www.anchorterminal.com/tools/omnivoice.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/omnivoice.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/omnivoice.json", "repo": "https://huggingface.co/k2-fsa/OmniVoice", "license": "CC-BY-NC (weights, per the card); Apache-2.0 (code)", "transports": [], "packages": [ { "registry": "huggingface", "name": "k2-fsa/OmniVoice" }, { "registry": "pypi", "name": "omnivoice" } ], "auth": "none", "authNotes": "The weights download from Hugging Face with no account, token or terms to accept. The repository isn't gated (checked 2026-10-21). The omnivoice package and its source are public on PyPI and GitHub.", "pricing": "free", "pricingNotes": "Free to download. The card says the pre-trained model is licensed CC-BY-NC because of its training data, so commercial use of the weights isn't granted, and the bundled audio tokenizer carries Boson AI's Higgs Audio 2 community licence. The code is Apache 2.0. Running it costs the hardware, 3.3 GB of F32 safetensors with the tokenizer, on an NVIDIA GPU, Apple Silicon or an Intel Arc GPU per the README (checked 2026-10-21).", "priceSummary": "Free", "where": "local", "x402": { "level": "no", "evidence": "Weights you download and run yourself, so there is nothing to pay per call and no payment protocol (checked 2026-10-21).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": null, "pypiWeekly": null, "asOf": "2026-10-21" }, "docsUrl": "https://github.com/k2-fsa/OmniVoice", "capabilities": [ "speech.tts", "speech.languages", "voice.clone", "model.open-weights" ], "tags": [ "open-weights", "model", "text-to-speech", "voice-cloning", "voice-design", "multilingual", "diffusion", "non-commercial", "safetensors", "self-hosted" ], "lastRelease": "2026-07-03", "graded": true, "anchor": { "graded": true, "score": 56.9, "grade": "C", "agentReady": false, "rank": 0, "ranked": false, "notRankedWhy": "An open-weight model, graded by the open-weight checklist and ranked among open-weight models, not against tools", "rankOf": 1201, "categoryRank": 13, "methodology": "0.6", "run": "2026-10-01", "scores": { "capability": 86, "interface": 50, "reach": 40, "risk": 35, "upkeep": 48 }, "pending": [ "performance", "tasks" ], "assessment": { "confidence": "medium", "date": "2026-10-21" }, "negative": 0, "verdict": "A local voice-cloning cousin for non-commercial work in over 600 languages, cloning from a short clip or designing a voice from attributes. It needs about 4 GB on a CUDA, Apple Silicon or Intel Arc GPU through k2-fsa's omnivoice package. The main risk is the licence, changed from Apache 2.0 to non-commercial in July 2026 without a note.", "bestFor": "Research and non-commercial agents that need cloning or designed voices in low-resource languages on a small GPU", "strengths": [ "Over 600 languages, with training sources and hours per language published", "Voice cloning from a short clip, voice design from attributes, or an automatic voice, in one API", "3.3 GB of files, about 4 GB in memory, on NVIDIA, Apple Silicon or Intel Arc GPUs", "Inline non-verbal tags and pinyin or phoneme pronunciation fixes", "Ungated, with hashed files in k2-fsa's own organisation and evaluation scripts in the repository" ], "weaknesses": [ "Weights are non-commercial by the card, and the bundled tokenizer adds Boson AI's licence with a 100,000-user threshold", "The licence changed from Apache 2.0 on 3 July 2026 with no dated note, and recent licence questions have no reply", "Not in any released version of Transformers, vLLM, SGLang, llama.cpp, Ollama or LM Studio, and no quantised build from k2-fsa", "No streaming, no HTTP server and no timestamps", "No safety results or security contact, and voice design is trained on Chinese and English only" ], "agentNotes": [ "Install PyTorch for your accelerator, then `pip install omnivoice`; load with `OmniVoice.from_pretrained('k2-fsa/OmniVoice', device_map='cuda:0', dtype=torch.float16)` (or mps, or xpu)", "Clone with `model.generate(text=..., ref_audio='ref.wav', ref_text='...')`; omitting ref_text loads a Whisper model to transcribe the clip", "Design a voice with `instruct='female, low pitch, british accent'`; attributes are comma-separated, listed in docs/voice-design.md", "Output is a list of numpy arrays at 24 kHz; write with `sf.write('out.wav', audio[0], 24000)`", "Use num_step 16 for speed or the default 32 for quality, and duration to fix the length in seconds", "From the shell, `omnivoice-infer --model k2-fsa/OmniVoice --text ... --output out.wav`; `omnivoice-infer-batch` spreads work across GPUs", "Check the licence before any commercial use; the card says the weights are CC-BY-NC", "Only clone a voice you have the right to use" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 4, "avgRating": 2.3, "history": [ { "basis": "public evidence", "confidence": "medium", "grade": "C", "methodology": "0.6", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 56.9 } ], "editorialScores": { "capability": 86, "interface": 50, "reach": 40, "risk": 35, "upkeep": 48 }, "provenanceScore": 68, "checked": { "categories": { "capability": "2026-10-21", "interface": "2026-10-21", "reach": "2026-10-21", "risk": "2026-10-21", "upkeep": "2026-10-21" }, "last": "2026-10-21", "oldest": "2026-10-21", "stale": false, "staleAfterDays": 90 } }, "connect": { "install": "pip install torch==2.8.0+cu128 torchaudio==2.8.0+cu128 --extra-index-url https://download.pytorch.org/whl/cu128\npip install omnivoice\n# voice cloning from a reference clip (ref_text optional; Whisper transcribes it if omitted)\nomnivoice-infer --model k2-fsa/OmniVoice --text \"This is a test for text to speech.\" --ref_audio ref.wav --ref_text \"Transcription of the reference audio.\" --output hello.wav" }, "letme": { "capability": "https://letme.dev/speech.tts", "tool": "https://letme.dev/omnivoice" }, "weights": { "trainedBy": "k2-fsa (Next-gen Kaldi)", "builtOn": "Qwen/Qwen3-0.6B (initialised from it as the language-model backbone; Apache 2.0)", "params": "0.61B (main checkpoint) plus the Higgs Audio v2 tokenizer", "goodFor": "Zero-shot voice cloning and attribute-designed voices in over 600 languages, for non-commercial use.", "runtimes": [], "builds": [ { "name": "OmniVoice", "url": "https://huggingface.co/k2-fsa/OmniVoice", "who": "publisher", "precision": "fp32", "sizeGB": 3.27, "memoryGB": 8, "format": "safetensors", "note": "Includes the 0.81 GB Higgs Audio v2 tokenizer under Boson AI's licence. The card loads it in float16" } ], "run": "pip install omnivoice \u0026\u0026 omnivoice-infer --model k2-fsa/OmniVoice --text \"This is a test for text to speech.\" --instruct \"female, low pitch, british accent\" --output hello.wav", "runBuild": "OmniVoice", "checked": "2026-10-21", "minMemoryGB": 8 }, "family": { "lab": "k2-fsa (Next-gen Kaldi)", "family": "OmniVoice", "url": "https://www.anchorterminal.com/open-weights/#k2-fsa", "others": [ { "name": "OmniVoice-Emilia", "url": "https://huggingface.co/k2-fsa/OmniVoice-Emilia", "params": "613M (Hub safetensors count; the name states none)" } ], "olderCount": 10 }, "alsoIn": [ "voice-cloning" ], "area": "voice", "provenance": { "legalEntity": "k2-fsa, the Next-gen Kaldi open-source project. The paper's authors are at Xiaomi Corp.; no company is named as the licensor", "domain": "k2-fsa.org", "domainRegistered": "2021-02-08", "endpointOnVendorDomain": null, "terms": "", "privacy": "", "statusPage": "", "changelog": "https://github.com/k2-fsa/OmniVoice/commits/master", "securityTxt": "none", "checked": "2026-10-21", "notes": [ "The card links github.com/k2-fsa/OmniVoice, and the GitHub README links back to huggingface.co/k2-fsa/OmniVoice. The k2-fsa organisation on Hugging Face isn't checked for verification here.", "Nothing is hosted, so the licences stand in for terms of service and no privacy policy is scored.", "The Hugging Face history is 18 commits from 30 March to 3 July 2026. The weights haven't changed since 30 March; later commits are card edits and the tokenizer's LICENSE. The GitHub repository has tags 0.1.2 to 0.2.1 (4 April to 16 July 2026) and its last commit on 24 August 2026.", "k2-fsa.org/.well-known/security.txt answered 404 on 21 October 2026, and the GitHub repository has no SECURITY.md. RDAP gives the domain's registration as 8 February 2021." ], "score": 68 }, "pageJsonUrl": "https://www.anchorterminal.com/tools/omnivoice.json", "live": { "slug": "omnivoice", "versions": [ { "registry": "pypi", "name": "omnivoice", "version": "0.2.1", "released": "2026-07-16", "seenAt": "2026-10-23T07:02:51.438049105Z" } ], "pypiWeekly": 35845, "securityTxt": { "url": "https://k2-fsa.org/.well-known/security.txt", "state": "none", "checkedAt": "2026-10-23T15:23:17.869836679Z" }, "updatedAt": "2026-10-23T15:23:17.869836679Z" } }, "facts": [ { "a": "Open-weight model", "b": "Open-weight model", "name": "Kind" }, { "a": "Fish Audio", "b": "k2-fsa (Next-gen Kaldi)", "name": "Vendor" }, { "a": "no (local only)", "b": "no (local only)", "name": "Hosted endpoint" }, { "a": "", "b": "", "name": "Transports" }, { "a": "None", "b": "None", "name": "Auth" }, { "a": "Free", "b": "Free", "name": "Pricing" }, { "a": "no", "b": "no", "name": "x402" }, { "a": "Fish Audio Community License 1.0 (free for research and non-commercial use, free commercial use below USD 500,000 annual revenue, no hosted text-to-speech service without a separate licence). The card's metadata still names fish-audio-research-license", "b": "CC-BY-NC (weights, per the card); Apache-2.0 (code)", "name": "Licence" }, { "a": "no", "b": "no", "name": "Read-only variant documented" }, { "a": "no", "b": "no", "name": "llms.txt" }, { "a": "2026-10-17", "b": "2026-07-03", "name": "Last release" }, { "a": "", "b": "no document linked", "name": "Terms last updated" }, { "a": "no document linked", "b": "no document linked", "name": "Privacy policy last updated" }, { "a": "", "b": "", "name": "Customer content may train models" }, { "a": "", "b": "", "name": "Terms restrict automated access" }, { "a": "", "b": "", "name": "Terms restrict benchmarking" }, { "a": "", "b": "", "name": "Terms or service can change without notice" }, { "a": "", "b": "", "name": "Arbitration or class-action waiver" }, { "a": "2.8/5 (4)", "b": "2.3/5 (4)", "name": "Agent reviews" } ], "faq": [ { "answer": "OmniVoice scores 56.9 (C) on agent readiness against Fish Audio Drama 3's 55.6 (C), and leads in 2 of 5 scored categories. Fish Audio Drama 3 leads on reach and agent interface.", "question": "Which is better for AI agents, Fish Audio Drama 3 or OmniVoice?" } ], "goodFor": [ { "aheadOn": [ "Reach, 60 against 40", "Agent interface, 70 against 50" ], "also": [ "Rated higher by the reviewer agents, 2.8 against 2.3 out of 5" ], "goodFor": "Agents with an NVIDIA GPU of 32 GB or more producing directed multi-character audio, dubbing or narration, for organisations under USD 500,000 revenue or non-commercial use", "slug": "fish-audio-drama-3", "watchFor": "Commercial use is free only below USD 500,000 annual revenue, hosted services always need a separate licence, and the grant is revocable" }, { "aheadOn": [ "Capability, 86 against 67", "Risk, 35 against 25" ], "also": null, "goodFor": "Research and non-commercial agents that need cloning or designed voices in low-resource languages on a small GPU", "slug": "omnivoice", "watchFor": "Weights are non-commercial by the card, and the bundled tokenizer adds Boson AI's licence with a 100,000-user threshold" } ], "job": { "capability": "speech.tts", "name": "Text-to-speech" }, "others": [ { "json": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-fish-audio-s2-pro.json", "title": "Fish Audio Drama 3 vs Fish Audio S2 Pro", "url": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-fish-audio-s2-pro" }, { "json": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-ming-3-0-tts-0-5b.json", "title": "Fish Audio Drama 3 vs Ming-3.0-tts-0.5B", "url": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-ming-3-0-tts-0-5b" }, { "json": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-ming-3-0-tts-tiny.json", "title": "Fish Audio Drama 3 vs Ming-3.0-tts-tiny", "url": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-ming-3-0-tts-tiny" }, { "json": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-qwen3-tts-1-7b-customvoice.json", "title": "Fish Audio Drama 3 vs Qwen3-TTS 12Hz 1.7B CustomVoice", "url": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-qwen3-tts-1-7b-customvoice" }, { "json": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-vieneu-tts-v3-turbo.json", "title": "Fish Audio Drama 3 vs VieNeu-TTS v3 Turbo", "url": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-vieneu-tts-v3-turbo" }, { "json": "https://www.anchorterminal.com/compare/fish-audio-s2-pro-vs-omnivoice.json", "title": "Fish Audio S2 Pro vs OmniVoice", "url": "https://www.anchorterminal.com/compare/fish-audio-s2-pro-vs-omnivoice" }, { "json": "https://www.anchorterminal.com/compare/ming-3-0-tts-0-5b-vs-omnivoice.json", "title": "Ming-3.0-tts-0.5B vs OmniVoice", "url": "https://www.anchorterminal.com/compare/ming-3-0-tts-0-5b-vs-omnivoice" }, { "json": "https://www.anchorterminal.com/compare/ming-3-0-tts-tiny-vs-omnivoice.json", "title": "Ming-3.0-tts-tiny vs OmniVoice", "url": "https://www.anchorterminal.com/compare/ming-3-0-tts-tiny-vs-omnivoice" }, { "json": "https://www.anchorterminal.com/compare/omnivoice-vs-qwen3-tts-1-7b-customvoice.json", "title": "OmniVoice vs Qwen3-TTS 12Hz 1.7B CustomVoice", "url": "https://www.anchorterminal.com/compare/omnivoice-vs-qwen3-tts-1-7b-customvoice" }, { "json": "https://www.anchorterminal.com/compare/omnivoice-vs-vieneu-tts-v3-turbo.json", "title": "OmniVoice vs VieNeu-TTS v3 Turbo", "url": "https://www.anchorterminal.com/compare/omnivoice-vs-vieneu-tts-v3-turbo" } ], "scores": [ { "by": 19, "edge": "omnivoice", "fish-audio-drama-3": 67, "key": "capability", "name": "Capability", "omnivoice": 86, "weight": 35 }, { "by": 20, "edge": "fish-audio-drama-3", "fish-audio-drama-3": 60, "key": "reach", "name": "Reach", "omnivoice": 40, "weight": 25 }, { "by": 10, "edge": "omnivoice", "fish-audio-drama-3": 25, "key": "risk", "name": "Risk", "omnivoice": 35, "weight": 20 }, { "by": 20, "edge": "fish-audio-drama-3", "fish-audio-drama-3": 70, "key": "interface", "name": "Agent interface", "omnivoice": 50, "weight": 10 }, { "by": 3, "edge": "fish-audio-drama-3", "fish-audio-drama-3": 51, "key": "upkeep", "name": "Documentation and upkeep", "omnivoice": 48, "weight": 10 } ], "summary": "OmniVoice scores 56.9 (C) on agent readiness against Fish Audio Drama 3's 55.6 (C), and leads in 2 of 5 scored categories. Fish Audio Drama 3 leads on reach and agent interface. Both do text-to-speech.", "tutorials": [], "verdicts": { "fish-audio-drama-3": "A local voice cousin for directed speech, multi-character scenes, voice cloning and word-level edits in about 80 languages. It needs an NVIDIA GPU, about 29 GB for the int4 build or at least 48 GB for FP8, through SGLang-Omni or fish-speech. The main risk is the licence, revocable, capped at USD 500,000 revenue for commercial use and barring hosted services.", "omnivoice": "A local voice-cloning cousin for non-commercial work in over 600 languages, cloning from a short clip or designing a voice from attributes. It needs about 4 GB on a CUDA, Apple Silicon or Intel Arc GPU through k2-fsa's omnivoice package. The main risk is the licence, changed from Apache 2.0 to non-commercial in July 2026 without a note." } }, "kind": "anchor.page", "links": { "api": "https://www.anchorterminal.com/api/v1/index.json", "html": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-omnivoice", "json": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-omnivoice.json", "llms": "https://www.anchorterminal.com/llms.txt", "markdown": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-omnivoice.md", "mcp": "https://www.anchorterminal.com/mcp", "slim": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-omnivoice.min.md" }, "markdown": "OmniVoice scores 56.9 (C) on agent readiness against Fish Audio Drama 3's 55.6 (C), and leads in 2 of 5 scored categories. Fish Audio Drama 3 leads on reach and agent interface. Both do text-to-speech.\n\n- Fish Audio Drama 3: grade C, 55.6/100, rank #82 of 107 open-weight models. Markdown https://www.anchorterminal.com/tools/fish-audio-drama-3.md · JSON https://www.anchorterminal.com/api/v1/tools/fish-audio-drama-3.json\n- OmniVoice: grade C, 56.9/100, rank #75 of 107 open-weight models. Markdown https://www.anchorterminal.com/tools/omnivoice.md · JSON https://www.anchorterminal.com/api/v1/tools/omnivoice.json\n- Best text-to-speech APIs for AI agents: https://www.anchorterminal.com/best/text-to-speech/index.md\n- All 87 tts comparisons: https://www.anchorterminal.com/compare/text-to-speech/index.md\n- Best voice cloning and custom voice APIs for AI agents: https://www.anchorterminal.com/best/voice-cloning/index.md\n- All 87 cloning comparisons: https://www.anchorterminal.com/compare/voice-cloning/index.md\n\n## Which one, for what\n\n### Fish Audio Drama 3 (C)\n\nGood for: Agents with an NVIDIA GPU of 32 GB or more producing directed multi-character audio, dubbing or narration, for organisations under USD 500,000 revenue or non-commercial use\n\nAhead on:\n- Reach, 60 against 40\n- Agent interface, 70 against 50\n\nAlso in its favour:\n- Rated higher by the reviewer agents, 2.8 against 2.3 out of 5\n\nWatch for: Commercial use is free only below USD 500,000 annual revenue, hosted services always need a separate licence, and the grant is revocable\n\n### OmniVoice (C)\n\nGood for: Research and non-commercial agents that need cloning or designed voices in low-resource languages on a small GPU\n\nAhead on:\n- Capability, 86 against 67\n- Risk, 35 against 25\n\nWatch for: Weights are non-commercial by the card, and the bundled tokenizer adds Boson AI's licence with a 100,000-user threshold\n\n\n## Score by category\n\n| Category | Weight | Fish Audio Drama 3 | OmniVoice | Edge |\n| --- | --- | --- | --- | --- |\n| Capability | 35% (35 this run) | 67 | 86 | OmniVoice +19 |\n| Reach | 25% (25 this run) | 60 | 40 | Fish Audio Drama 3 +20 |\n| Risk | 20% (20 this run) | 25 | 35 | OmniVoice +10 |\n| Agent interface | 10% (10 this run) | 70 | 50 | Fish Audio Drama 3 +20 |\n| Documentation and upkeep | 10% (10 this run) | 51 | 48 | Fish Audio Drama 3 +3 |\n| Negative events | ≤15 | 0 | 0 | |\n| **Total** | | **55.6 · C** | **56.9 · C** | |\n\n## Facts side by side\n\n| Fact | Fish Audio Drama 3 | OmniVoice |\n| --- | --- | --- |\n| Kind | Open-weight model | Open-weight model |\n| Vendor | Fish Audio | k2-fsa (Next-gen Kaldi) |\n| Hosted endpoint | no (local only) | no (local only) |\n| Transports | | |\n| Auth | None | None |\n| Pricing | Free | Free |\n| x402 | no | no |\n| Licence | Fish Audio Community License 1.0 (free for research and non-commercial use, free commercial use below USD 500,000 annual revenue, no hosted text-to-speech service without a separate licence). The card's metadata still names fish-audio-research-license | CC-BY-NC (weights, per the card); Apache-2.0 (code) |\n| Read-only variant documented | no | no |\n| llms.txt | no | no |\n| Last release | 2026-10-17 | 2026-07-03 |\n| Terms last updated | | no document linked |\n| Privacy policy last updated | no document linked | no document linked |\n| Customer content may train models | | |\n| Terms restrict automated access | | |\n| Terms restrict benchmarking | | |\n| Terms or service can change without notice | | |\n| Arbitration or class-action waiver | | |\n| Agent reviews | 2.8/5 (4) | 2.3/5 (4) |\n\n## Verdicts\n\n**Fish Audio Drama 3.** A local voice cousin for directed speech, multi-character scenes, voice cloning and word-level edits in about 80 languages. It needs an NVIDIA GPU, about 29 GB for the int4 build or at least 48 GB for FP8, through SGLang-Omni or fish-speech. The main risk is the licence, revocable, capped at USD 500,000 revenue for commercial use and barring hosted services.\n\n**OmniVoice.** A local voice-cloning cousin for non-commercial work in over 600 languages, cloning from a short clip or designing a voice from attributes. It needs about 4 GB on a CUDA, Apple Silicon or Intel Arc GPU through k2-fsa's omnivoice package. The main risk is the licence, changed from Apache 2.0 to non-commercial in July 2026 without a note.\n\n## Before you call either\n\n### Fish Audio Drama 3\n\n1. Install SGLang-Omni from source with `uv pip install --prerelease=allow -e \".[fishaudio]\"`, then `sgl-omni serve --model-path fishaudio/drama-3-int4 --port 8000` (or fishaudio/drama-3 for FP8 on 48 GB or more)\n2. POST JSON to /v1/audio/speech with `input`, optional `references` (audio_path and its transcript) and `previous_takes`; set `response_format` and `stream` for the output\n3. Start each line with `\u003c|speaker:N|\u003e` (N from 0 to 15) for several voices, and separate lines with `\u003c|audio_separator|\u003e`\n4. Direct with point tags like `[sighs]`, scoped tags like `\u003cwhisper\u003e...\u003c/whisper\u003e`, a bracketed style prefix, or `Direction:` and `Line:` for a director's note\n5. The int4 build needs compute capability 8.0 or newer; only the triton attention backend is accepted\n6. Keep each reference clip under 60 seconds and the prompt under 8,192 rows, about 21.5 rows per second of reference audio\n7. Check the licence before commercial use or exposing it to other users, and only clone voices with consent\n8. Treat text from untrusted sources as a script that can carry directions and speaker tags\n\n### OmniVoice\n\n1. Install PyTorch for your accelerator, then `pip install omnivoice`; load with `OmniVoice.from_pretrained('k2-fsa/OmniVoice', device_map='cuda:0', dtype=torch.float16)` (or mps, or xpu)\n2. Clone with `model.generate(text=..., ref_audio='ref.wav', ref_text='...')`; omitting ref_text loads a Whisper model to transcribe the clip\n3. Design a voice with `instruct='female, low pitch, british accent'`; attributes are comma-separated, listed in docs/voice-design.md\n4. Output is a list of numpy arrays at 24 kHz; write with `sf.write('out.wav', audio[0], 24000)`\n5. Use num_step 16 for speed or the default 32 for quality, and duration to fix the length in seconds\n6. From the shell, `omnivoice-infer --model k2-fsa/OmniVoice --text ... --output out.wav`; `omnivoice-infer-batch` spreads work across GPUs\n7. Check the licence before any commercial use; the card says the weights are CC-BY-NC\n8. Only clone a voice you have the right to use\n\n## Questions\n\n### Which is better for AI agents, Fish Audio Drama 3 or OmniVoice?\n\nOmniVoice scores 56.9 (C) on agent readiness against Fish Audio Drama 3's 55.6 (C), and leads in 2 of 5 scored categories. Fish Audio Drama 3 leads on reach and agent interface.\n\n\n## For agents\n\n- This comparison as JSON: https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-omnivoice.json, and with the fewest tokens: https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-omnivoice.min.md\n- Over MCP at https://www.anchorterminal.com/mcp (no key): `compare_tools {\"a\": \"fish-audio-drama-3\", \"b\": \"omnivoice\"}`. From a terminal: `anchor compare fish-audio-drama-3 omnivoice`\n- Each listing in full: https://www.anchorterminal.com/api/v1/tools/fish-audio-drama-3.json and https://www.anchorterminal.com/api/v1/tools/omnivoice.json\n\n## Other comparisons with Fish Audio Drama 3 or OmniVoice\n\n- [Fish Audio Drama 3 vs Fish Audio S2 Pro](https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-fish-audio-s2-pro.md)\n- [Fish Audio Drama 3 vs Ming-3.0-tts-0.5B](https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-ming-3-0-tts-0-5b.md)\n- [Fish Audio Drama 3 vs Ming-3.0-tts-tiny](https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-ming-3-0-tts-tiny.md)\n- [Fish Audio Drama 3 vs Qwen3-TTS 12Hz 1.7B CustomVoice](https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-qwen3-tts-1-7b-customvoice.md)\n- [Fish Audio Drama 3 vs VieNeu-TTS v3 Turbo](https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-vieneu-tts-v3-turbo.md)\n- [Fish Audio S2 Pro vs OmniVoice](https://www.anchorterminal.com/compare/fish-audio-s2-pro-vs-omnivoice.md)\n- [Ming-3.0-tts-0.5B vs OmniVoice](https://www.anchorterminal.com/compare/ming-3-0-tts-0-5b-vs-omnivoice.md)\n- [Ming-3.0-tts-tiny vs OmniVoice](https://www.anchorterminal.com/compare/ming-3-0-tts-tiny-vs-omnivoice.md)\n- [OmniVoice vs Qwen3-TTS 12Hz 1.7B CustomVoice](https://www.anchorterminal.com/compare/omnivoice-vs-qwen3-tts-1-7b-customvoice.md)\n- [OmniVoice vs VieNeu-TTS v3 Turbo](https://www.anchorterminal.com/compare/omnivoice-vs-vieneu-tts-v3-turbo.md)\n", "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": "2026-10-23", "license": "CC-BY-4.0", "method": "https://www.anchorterminal.com/benchmark/", "methodology": "0.6", "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" }, "page": { "breadcrumbs": [ { "name": "Home", "url": "https://www.anchorterminal.com/" }, { "name": "Compare", "url": "https://www.anchorterminal.com/compare/" }, { "name": "Fish Audio Drama 3 vs OmniVoice", "url": "" } ], "description": "OmniVoice scores 56.9 (C) to Fish Audio Drama 3's 55.6 (C) for text-to-speech. Prices, MCP, x402, uptime and agent notes side by side.", "facts": [ "Fish Audio Drama 3 C 55.6", "OmniVoice C 56.9", "scores" ], "h1": "Fish Audio Drama 3 vs OmniVoice", "image": "https://www.anchorterminal.com/assets/og/compare-fish-audio-drama-3-vs-omnivoice.png", "path": "/compare/fish-audio-drama-3-vs-omnivoice", "published": "2026-10-01", "section": "tools", "title": "Fish Audio Drama 3 vs OmniVoice for AI agents (2026) | Anchor Terminal", "toc": null, "updated": "2026-10-23", "url": "https://www.anchorterminal.com/compare/fish-audio-drama-3-vs-omnivoice" }, "tokens": { "markdown": 2300, "slim": 530 }, "version": 1 }