G
GenAI Hub
🎙️
🎵 Voice & Audio

Hume AI (EVI)

by Hume AI

Hume AI's Empathic Voice Interface (EVI) is a voice AI API that understands and responds to human emotional expression in real time, enabling developers to build emotionally intelligent voice applications.

voice aiemotion aitext-to-speechspeech-to-speechapireal-time
Open ToolDocumentation

Quick Start

  1. Get API credentials — Sign up at platform.hume.ai and create an API key and Secret Key from the dashboard.
  2. Install the SDK — Run pip install hume (Python) or npm install @humeai/sdk (JavaScript/TypeScript).
  3. Open a WebSocket session — Connect to the EVI WebSocket endpoint using your credentials; EVI uses a persistent socket for real-time bidirectional audio streaming.
  4. Stream audio in and out — Send raw audio bytes (PCM 16-bit) from the microphone; receive audio bytes back along with JSON messages containing transcripts and emotion scores.
  5. Handle events — Parse incoming messages for types like user_message, assistant_message, and audio_output to drive your UI and play back the response.
Full quickstart guide

How-To Guides

2
1. Install: pip install hume websockets. 2. Obtain HUME_API_KEY and HUME_SECRET_KEY from platform.hume.ai. 3. Generate a short-lived access token via the REST token endpoint using your key pair. 4. Open a WebSocket to wss://api.hume.ai/v0/evi/chat?access_token=TOKEN. 5. Capture microphone audio using pyaudio or sounddevice in PCM 16kHz mono format. 6. Send audio chunks as binary WebSocket frames. 7. Receive JSON messages — when type is 'audio_output', base64-decode the audio and play it back. 8. Listen for emotion scores in 'user_message' events to adapt your application UI (e.g., show empathy indicators).

Tool Info

Company
Hume AI
Category
🎵 Voice & Audio
How-to Guides
2
Official websiteDocumentationQuickstart guide

Tags

voice aiemotion aitext-to-speechspeech-to-speechapireal-time

Related Tools

♪
audio

ElevenLabs

ElevenLabs

The leading AI voice platform. Clone voices from a short audio sample, generate hyper-realistic speech, and produce multilingual audio. Widely used in content creation and apps.

voice cloningTTSmultilingual
View details →Featured
W
audio

OpenAI Whisper

OpenAI

OpenAI's open-source speech recognition model. Highly accurate transcription in 99 languages. Runs locally, available via the OpenAI API, and powering many apps.

transcriptionopen-sourcemultilingual
View details →
▶
audio

PlayHT

PlayHT

AI voice generation platform with 900+ voices in 142 languages. Offers voice cloning, ultra-realistic TTS, and an API for building voice-enabled applications.

TTSvoice cloning900+ voices
View details →
♫
audio

Suno AI

Suno

Create full songs with lyrics, vocals, and instrumentation from a text prompt. Produces studio-quality music in any genre in seconds. No musical knowledge required.

music generationlyricsvocals
View details →Featured
Back to all tools