| { |
| "manifest_version": "1.0", |
| "kind": "model-capability-reference", |
| "documentation_reviewed_at": "2026-11-03", |
| "model": { |
| "name": "Maya 2 Native", |
| "aliases": [ |
| "Maya2 Native", |
| "Maya-2-Native", |
| "maya2-native" |
| ], |
| "publisher": "Maya Research", |
| "repository_id": "maya-research/maya2-native", |
| "url": "https://huggingface.co/maya-research/maya2-native", |
| "task": "text-to-speech", |
| "release": "1.0", |
| "checkpoint_revision_reviewed": "76aedafc325fb72413569a0e15121d58ea2ac312", |
| "unique_parameter_count": 2502398656, |
| "stored_tensor_bytes": 5004797312, |
| "weight_dtype": "bfloat16", |
| "languages": [ |
| { |
| "code": "as", |
| "name": "Assamese" |
| }, |
| { |
| "code": "bn", |
| "name": "Bengali" |
| }, |
| { |
| "code": "en", |
| "name": "English" |
| }, |
| { |
| "code": "gu", |
| "name": "Gujarati" |
| }, |
| { |
| "code": "hi", |
| "name": "Hindi" |
| }, |
| { |
| "code": "kn", |
| "name": "Kannada" |
| }, |
| { |
| "code": "ml", |
| "name": "Malayalam" |
| }, |
| { |
| "code": "mr", |
| "name": "Marathi" |
| }, |
| { |
| "code": "or", |
| "name": "Odia" |
| }, |
| { |
| "code": "pa", |
| "name": "Punjabi" |
| }, |
| { |
| "code": "ta", |
| "name": "Tamil" |
| }, |
| { |
| "code": "te", |
| "name": "Telugu" |
| } |
| ], |
| "capabilities": { |
| "streaming_audio_output": true, |
| "built_in_voices": true, |
| "zero_shot_voice_cloning": true, |
| "cross_lingual_voice_cloning": true, |
| "bundled_text_normalization": true |
| }, |
| "audio": { |
| "sample_rate_hz": 24000, |
| "channels": 1, |
| "model_frame_duration_ms": 80, |
| "generate_numpy_dtype": "float32", |
| "stream_numpy_dtype": "int16", |
| "raw_pcm_encoding": "pcm_s16le" |
| }, |
| "voices": [ |
| { |
| "id": "Ahana", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "pa", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Vaidehi", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "pa", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Malini", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "pa", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Charvi", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "pa", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Jahnavi", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Anirban", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "pa", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Nakul", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "pa", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Vasant", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Siddhant", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "pa", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Satvik", |
| "training_language_codes_at_least_one_hour": [ |
| "bn", |
| "en", |
| "gu", |
| "hi", |
| "kn", |
| "ml", |
| "mr", |
| "or", |
| "pa", |
| "ta", |
| "te" |
| ] |
| }, |
| { |
| "id": "Abhay", |
| "training_language_codes_at_least_one_hour": [ |
| "en", |
| "hi" |
| ] |
| }, |
| { |
| "id": "Nivedita", |
| "training_language_codes_at_least_one_hour": [ |
| "en", |
| "hi" |
| ] |
| }, |
| { |
| "id": "Ishwari", |
| "training_language_codes_at_least_one_hour": [ |
| "hi" |
| ] |
| }, |
| { |
| "id": "Ritvik", |
| "training_language_codes_at_least_one_hour": [ |
| "hi" |
| ] |
| }, |
| { |
| "id": "Vrinda", |
| "training_language_codes_at_least_one_hour": [ |
| "hi" |
| ] |
| }, |
| { |
| "id": "Sreekar", |
| "training_language_codes_at_least_one_hour": [ |
| "en", |
| "hi", |
| "te" |
| ] |
| }, |
| { |
| "id": "Kayal", |
| "training_language_codes_at_least_one_hour": [ |
| "en", |
| "ta" |
| ] |
| }, |
| { |
| "id": "Shaunak", |
| "training_language_codes_at_least_one_hour": [ |
| "en", |
| "hi" |
| ] |
| }, |
| { |
| "id": "Akshat", |
| "training_language_codes_at_least_one_hour": [ |
| "en", |
| "hi" |
| ] |
| }, |
| { |
| "id": "Tanushree", |
| "training_language_codes_at_least_one_hour": [ |
| "en", |
| "hi" |
| ] |
| } |
| ], |
| "voice_coverage_note": "Voices can be requested in any supported language. Unlisted voice-language pairs are cross-lingual; no built-in voice has at least one hour of Assamese training data." |
| }, |
| "runtime": { |
| "package": "maya3", |
| "python_import": "from maya3 import Maya3TTS", |
| "loader": "Maya3TTS.from_pretrained", |
| "checkpoint_config_filename": "maya3_config.json", |
| "documented_os": [ |
| "Linux" |
| ], |
| "documented_python_versions": [ |
| "3.10", |
| "3.11", |
| "3.12" |
| ], |
| "gpu": "NVIDIA", |
| "documented_cuda_version": "12.8", |
| "suggested_starting_max_seats": 1, |
| "reported_single_seat_loading_peak_gb_excluding_cuda_context": 9.2, |
| "recommended_starting_gpu_memory_gb_or_more": 12, |
| "memory_source": "https://huggingface.co/maya-research/maya2-native#hardware-and-memory", |
| "normalization_defaults": { |
| "python": false, |
| "local_http": true, |
| "local_websocket": true |
| }, |
| "offline_dependencies": [ |
| "maya-research/maya2-native", |
| "kyutai/mimi", |
| "speechbrain/spkrec-ecapa-voxceleb" |
| ], |
| "generic_transformers_pipeline_documented": false |
| }, |
| "local_interfaces": { |
| "python": { |
| "generate": "float32 waveform", |
| "stream": "iterator of int16 PCM arrays", |
| "astream": "async iterator of int16 PCM arrays", |
| "language_parameter": "lang" |
| }, |
| "http": { |
| "method": "POST", |
| "path": "/v1/audio/speech", |
| "language_parameter": "language", |
| "streaming_formats": [ |
| "pcm", |
| "wav", |
| "mp3", |
| "opus" |
| ], |
| "completed_formats": [ |
| "flac" |
| ], |
| "example_request": { |
| "model": "maya3", |
| "input": "नमस्ते।", |
| "language": "hi", |
| "voice": "Ahana", |
| "response_format": "wav", |
| "normalize": true |
| }, |
| "supported_speed": [ |
| 1.0 |
| ] |
| }, |
| "websocket": { |
| "path": "/v1/tts/stream", |
| "audio_encoding": "pcm_s16le", |
| "sample_rate_hz": 24000, |
| "request_contexts": true, |
| "cancellation": true, |
| "protocol_source": "https://raw.githubusercontent.com/MayaResearch/maya2-native/main/maya3/server.py" |
| }, |
| "health": { |
| "method": "GET", |
| "path": "/health", |
| "ready_status": 200, |
| "loading_status": 503 |
| }, |
| "hosted_api_contract_equivalent": false |
| }, |
| "licences": { |
| "weights": "Apache-2.0", |
| "inference_code": "Apache-2.0", |
| "components": [ |
| { |
| "id": "Qwen/Qwen3-1.7B-Base", |
| "licence": "Apache-2.0" |
| }, |
| { |
| "id": "kyutai/mimi", |
| "licence": "CC-BY-4.0" |
| }, |
| { |
| "id": "speechbrain/spkrec-ecapa-voxceleb", |
| "licence": "Apache-2.0" |
| } |
| ] |
| }, |
| "references": { |
| "model_card_markdown": "https://huggingface.co/maya-research/maya2-native/resolve/main/README.md", |
| "agent_guide": "https://huggingface.co/maya-research/maya2-native/resolve/main/AGENT_GUIDE.md", |
| "llms_index": "https://huggingface.co/maya-research/maya2-native/resolve/main/llms.txt", |
| "voice_metadata": "https://huggingface.co/maya-research/maya2-native/resolve/main/voices/voices.json", |
| "checkpoint_configuration": "https://huggingface.co/maya-research/maya2-native/resolve/main/maya3_config.json", |
| "inference_code": "https://github.com/MayaResearch/maya2-native", |
| "engine_documentation": "https://raw.githubusercontent.com/MayaResearch/maya2-native/main/docs/ENGINE.md", |
| "normalization_documentation": "https://raw.githubusercontent.com/MayaResearch/maya2-native/main/maya3/text/README.md", |
| "benchmark_details": "https://huggingface.co/maya-research/maya2-native#evaluation", |
| "external_leaderboard": "https://voicearena.com/tts-leaderboard/hindi" |
| }, |
| "scope_notes": [ |
| "Descriptive model manifest; not an executable MCP tool or standard plugin registration.", |
| "Audio-frame duration is not time to first audio.", |
| "Model-card benchmarks are scoped to their documented hardware, evaluation and date.", |
| "The reference engine is a text-to-speech stage; ASR, dialogue, telephony and playback turn management are separate components." |
| ] |
| } |
|
|
