Summary: Hume AI is a research lab and technology company focused on emotionally intelligent AI — described on its site as building “AI that serves human goals and emotional well‑being.” Primary products and developer offerings described across hume.ai and dev.hume.ai include Octave (an expressive TTS / speech‑language voice model), EVI (Empathic Voice Interface for real‑time, emotion‑aware voice interaction), and Expression Measurement (multimodal measurement of voice, face, and language). The site emphasizes research and an evidence base and references publications and research work. Key capabilities: Octave is described as a voice‑based LLM that uses context to produce humanlike emotion, cadence, and nuance; it supports natural‑language acting instructions (e.g., “sound sarcastic”), voice cloning and voice design from prompts or short recordings, an Instant Mode/streaming low‑latency option (time‑to‑first‑token around ~200 ms reported for Octave 2 in the sources referenced), and multi‑speaker/multi‑character workflows suitable for audiobooks and studio‑quality podcasts. EVI (Empathic Voice Interface) is described as a real‑time voice interaction system that measures user prosody and other signals and responds with expressive speech, end‑of‑turn detection, interruptibility, and improved conversational EQ; use cases include coaching, interviewing, digital companions, and real‑time agents. Expression Measurement offers models that capture dimensions of expression across voice, face, and language (facial expressions, prosody, transcription, semantic/emotional metrics), available as streaming and batch endpoints with pay‑as‑you‑go billing for minutes/images. Developer experience: multi‑language support reported for 11+ languages (examples listed on the site include English, Japanese, Korean, Spanish, French, Portuguese, Italian, German, Russian, Hindi, Arabic). SDKs and quickstarts are provided for Python, TypeScript, Swift, React, and .NET, plus CLI and Node/Python quickstarts; developer ergonomics include playground(s) / no‑code testing for TTS, EVI, and Expression Measurement, documentation and guides (acting instructions, continuation across utterances, voice library), and examples for streaming vs non‑streaming endpoints and how to save voices. Pricing: tiered plan structure (Free → Starter → Creator → Pro → Scale → Business/Enterprise) plus pay‑as‑you‑go Expression Measurement billing; plans scale by monthly character/TTS minute quotas, RPM (requests per minute), number of projects, team seats, and availability of features such as voice cloning permissions, commercial licenses, concurrent connections, and external LLM support. The Pricing page and blog posts are referenced for specific numeric examples (the user-provided notes include a Starter example shown as $3/month and mentions example per‑minute EVI pricing shown in blog posts as low as $0.02/min at higher volumes; the sources also note Octave 2 as materially cheaper than Octave 1). Enterprise: higher tiers / custom contracts mention SOC 2 Type II, GDPR, and HIPAA compliance options. Timeline and launches (as referenced on the site/blog): Octave introduced in late 2024/2025 blog series; Octave 2 launch noted on Oct 1, 2025 (reported as lower latency and cost improvements); ongoing EVI releases/iterations (EVI 2 / EVI 3 / EVI 4‑mini referenced); a blog entry noted Nov 7, 2025 describing expanded persona/voice generation and limited early access for safety evaluation. Sources visited (primary): Home (https://hume.ai/), Pricing (https://www.hume.ai/pricing), Text‑to‑speech (https://www.hume.ai/text-to-speech), Developer intro/docs (https://dev.hume.ai/intro and TTS docs), About (https://www.hume.ai/about), and blog posts including Introducing OCTAVE (https://www.hume.ai/blog/introducing-octave) and Octave 2 launch (https://www.hume.ai/blog/octave-2-launch). Notes on accuracy and provenance: this entry is compiled from the pages and blog posts listed in external_links. Where the original notes reported examples or approximations (e.g., Starter example shown as $3/month, reported per‑minute EVI pricing examples), those values are preserved as reported from the referenced pages/posts; for exact, current numeric limits or promotional offers, consult the Pricing page or contact sales as recommended by the source material.