{"service":"CognisArk TTS (tts1) — Fish Audio S2-Pro","tagline":"The Ark of Knowledge · Serverless GPU Voice Studio","base_url":"https://tts1.cognisark.com","studio_ui":"https://tts1.cognisark.com/","license":{"model":"fishaudio/s2-pro","weights":"CC-BY-NC-SA-4.0","production_review_required":true,"note":"Non-commercial evaluation & internal research. Review commercial licensing before using in paid products."},"backend":{"provider":"runpod","gpu":"NVIDIA RTX A5000 24GB (AMPERE_24)","endpoint_id":"yqiecs97jd9bm8","model":"Fish Audio S2-Pro (Dual-AR Acoustic Foundation Model)","architecture":"Autoregressive LLM semantic transformer + DAC audio codec decoder","queue":"RunPod Serverless Queue — Cloudflare Worker front door"},"auth":{"header":"Authorization: Bearer <TTS1_API_KEY>","alternate_header":"X-Api-Key: <TTS1_API_KEY>","required_for":["/run","/status/:id","/api/v1/tts","/api/v1/tts/bulk"],"public":["/","/studio","/health","/help","/voices"]},"available_voices":[{"id":"gemini_puck","name":"Puck","gender":"male","style":"Playful, upbeat, youthful and expressive","cloned_from":"Gemini Puck","recommended_for":"Social media, energetic hooks, conversational dialogue"},{"id":"gemini_charon","name":"Charon","gender":"male","style":"Deep, calm, authoritative and articulate","cloned_from":"Gemini Charon","recommended_for":"Documentaries, cinematic narration, serious explainers"},{"id":"gemini_aoede","name":"Aoede","gender":"female","style":"Bright, brisk, clear and engaging","cloned_from":"Gemini Aoede","recommended_for":"Product tours, educational content, keynote intros"},{"id":"gemini_laomedeia","name":"Laomedeia","gender":"female","style":"Warm, soothing, gentle and friendly","cloned_from":"Gemini Laomedeia","recommended_for":"Customer care, meditation, onboarding, storybooks"},{"id":"gemini_achernar","name":"Achernar","gender":"male","style":"Confident, articulate and professional","cloned_from":"Gemini Achernar","recommended_for":"Corporate presentations, executive briefings, tutorials"},{"id":null,"name":"Base Neural","gender":"neutral","style":"Balanced neutral baseline narrator","cloned_from":"Fish Audio S2-Pro Base","recommended_for":"General purpose zero-prompt narration"}],"parameters":{"text":{"type":"string","required":true,"desc":"Text to synthesize (plain text or emotion tags)"},"reference_id":{"type":"string | null","required":false,"desc":"Preset voice ID (e.g. 'gemini_puck', 'gemini_charon', or null for base)"},"format":{"type":"string","default":"wav","options":["wav","pcm","mp3","opus"],"desc":"Audio output format"},"temperature":{"type":"float","default":0.7,"range":"0.1 - 1.0","desc":"Controls speech expressiveness and pitch variance"},"top_p":{"type":"float","default":0.8,"range":"0.1 - 1.0","desc":"Nucleus sampling threshold"},"repetition_penalty":{"type":"float","default":1.1,"range":"0.9 - 2.0","desc":"Penalizes repeated phonemes"},"chunk_length":{"type":"int","default":300,"range":"100 - 500","desc":"Max text token chunk size before autoregressive split"},"references":{"type":"array","required":false,"desc":"Dynamic inline voice cloning: [{ audio: '<base64_wav>', text: '<transcript>' }]"}},"endpoints":[{"method":"GET","path":"/","auth":false,"desc":"Interactive CognisArk Voice & Script Studio web application"},{"method":"GET","path":"/health","auth":false,"desc":"Liveness and RunPod GPU connectivity check"},{"method":"GET","path":"/help","auth":false,"desc":"API documentation and schema specification"},{"method":"GET","path":"/voices","auth":false,"desc":"List all built-in cloned voice presets"},{"method":"POST","path":"/api/v1/tts","auth":true,"desc":"Enqueue a single speech synthesis job","example_body":{"text":"Welcome to CognisArk Video Factory.","reference_id":"gemini_puck","format":"wav"},"response":{"id":"job_id_string","status":"IN_QUEUE"}},{"method":"GET","path":"/status/:id","auth":true,"desc":"Check job status. Terminal status: COMPLETED | FAILED","response":{"id":"job_id_string","status":"COMPLETED","output":{"audio_base64":"UklGRi...","bytes":1466412,"elapsed_sec":22.18}}},{"method":"POST","path":"/api/v1/tts/bulk","auth":true,"desc":"Enqueue batch synthesis jobs (max 32 chunks)","example_body":{"items":[{"text":"Section 1 narration","reference_id":"gemini_charon"},{"text":"Section 2 narration","reference_id":"gemini_aoede"}]}},{"method":"POST","path":"/run","auth":true,"desc":"Raw RunPod-compatible queue endpoint accepting { input: { ... } }"}],"pricing_and_performance":{"compute_rate":"$0.69/hr on RunPod Serverless GPU (~$0.00031/sec)","effective_cost":"~$7.00 – $10.00 per 1M characters (~$0.45 – $0.65 per hour of generated audio)","comparison":"95% lower cost than ElevenLabs ($150–$300/1M chars); 50% lower than standard API providers","idle_cost":"$0.00 (scale-to-zero serverless worker)"},"quick_start":{"studio":"Visit https://tts1.cognisark.com/ in your browser for the visual script editor.","curl_synthesis":"curl -s -X POST https://tts1.cognisark.com/api/v1/tts   -H \"Authorization: Bearer $TTS1_API_KEY\"   -H \"Content-Type: application/json\"   -d '{\"text\": \"[excited] Welcome to CognisArk!\", \"reference_id\": \"gemini_puck\"}'","curl_poll":"curl -s https://tts1.cognisark.com/status/<JOB_ID> -H \"Authorization: Bearer $TTS1_API_KEY\""}}