Replacing Hume AI’s Expression Measurement API? Here’s What Imentiv AI Offers

Hume AI's Expression Measurement API is going to sunset. If you've built a product on Hume AI's Expression Measurement API whether for UX research, recruitment, mental wellness, ad testing, or customer insight, the clock is ticking. You have until June 14, 2026 before API access is cut off and job results can no longer be downloaded.

The good news: If you're now looking for a Hume AI Expression Measurement API alternative, you don’t have to rebuild from scratch. Imentiv AI offers a production-ready, Multimodal Emotion Analysis APIs that maps to Hume AI's Expression Measurement core capabilities and, in several areas, goes further. The emotion API is live, documented, and available to start using today.

 

What Imentiv AI Offers

Imentiv AI is a multimodal emotion analysis platform that lets you detect and measure human emotional states from video, audio, text, and images via a clean REST APIs. It returns structured JSON with emotion scores, valence, arousal, and temporal breakdowns.

• Video Emotion API

Face-by-face, frame-by-frame emotion analysis from video files or YouTube URLs. Includes personality trait analysis via the OCEAN model.

 

• Audio Emotion API

Voice tone, pitch, and speech pattern analysis with speaker diarization and track each speaker's emotions separately.

 

• Text Emotion API

Sentence-by-sentence analysis detecting 32 distinct emotions from raw text.

 

• Image Emotion API

Detects facial expressions and emotional states from static images, with per-face bounding boxes.   

 

Hume AI vs Imentiv AI

A direct comparison of what each platform offers across the dimensions that matter most for a migration decision.

 

"Will I lose the core features I depend on?"

This is the question every migration starts with. Here's how Hume AI's core capabilities map to Imentiv AI's equivalents and where you get something extra.

Face → Face

Hume AI

Face model — Facial expressions are modeled as movement-based signals derived from facial action coding systems.

 

Imentiv AI

Video and Image Emotion APIs — Expressions are interpreted as primary emotions with bounding boxes, valence, arousal, and frame-level tracking.

Speech Prosody → Audio Emotion

Hume AI

Speech prosody model — Speech is analyzed through prosody and vocal bursts, capturing detailed acoustic variations.

 

Imentiv AI

Audio Emotion API — Emotional shifts are derived from tone, pitch, and speech patterns with speaker-level segmentation.

 

Language Model → Text Emotion

Hume AI

Language model — Language-based modeling captures emotional and tonal signals from text.

 

 

Imentiv AI

Text Emotion API — Sentences are mapped to clear emotional labels along with mood, valence, and dominant emotion.

 

Multimodal Jobs → Multimodal API

Hume AI

Run multiple models (face, prosody, language) on the same file in one batch job.

 

Imentiv AI

The Video Emotion API combines facial, vocal, and speech content analysis in a single call, multimodal by default.

What you gain as a bonus

Imentiv AI offers capabilities that Hume AI's Expression Measurement API did not include: Big Five OCEAN personality profiling from video content, native YouTube URL support, built-in speaker diarization for multi-speaker audio, and an Emotion Graph that visualizes emotional trajectories over time.

 

How to Get Started with Imentiv AI

• Create an account

Sign up at  imentiv.ai.  Once registered, navigate to My Account → My Profile to retrieve your API key. This key authenticates all API requests.

• Choose your endpoint

Imentiv AI provides four REST endpoints depending on your input type.

• Upload your media and call the API

Send a POST request with your file and API key. Imentiv AI returns structured JSON with emotion scores, valence, arousal, and timeline data.

• Parse and use the results

The response format is consistent across modalities. Adapt your existing data pipeline by remapping Hume's expression dimension keys to Imentiv AI's emotion labels and valence/arousal scores.

 

API Endpoints

POST https://api.imentiv.ai/v2/videos Video Emotion Analysis

POST https://api.imentiv.ai/v2/audios Audio Emotion Analysis

POST https://api.imentiv.ai/v2/texts Text Emotion Analysis

POST https://api.imentiv.ai/v2/images Image Emotion Analysis  

 

Migrating a Common Use Case: Facial Emotion from Video

One of the most common Hume’s Expression Measurement API use cases is analyzing facial emotion from a video file. Here's what that looked like on Hume AI, and how you'd do the same thing on Imentiv AI.

Hume AI

import os

import time

from hume import HumeClient

from hume.expression_measurement.batch import Face, Models

# 1. Initialize Client

client = HumeClient(api_key=os.environ.get("HUME_API_KEY"))

# 2. Start Job

job_id_response = client.expression_measurement.batch.start_inference_job(

    models=Models(face=Face()),

    urls=["https://example.com/interview.mp4"],

)

job_id = job_id_response.body

# 3. Poll for Completion

while True:

    job_details = client.expression_measurement.batch.get_job_details(id=job_id)

    status = job_details.body.state.status

    if status == "COMPLETED":

       print("Job completed.")

        break

    elif status == "FAILED":

        raise RuntimeError(f"Hume job failed: {job_details.body}")

    time.sleep(3)

# 4. Fetch and Process Predictions

predictions_response = client.expression_measurement.batch.get_job_predictions(id=job_id)

predictions = predictions_response.body

 

for result in predictions:

    for prediction in result.results.predictions:

        face_model = prediction.models.face

        if face_model is None:

            continue

        for group in face_model.grouped_predictions:

            for pred in group.predictions:

                top_emotion = max(pred.emotions, key=lambda e: e.score)

                print(f"Top emotion: {top_emotion.name} ({top_emotion.score:.3f})")

 

Imentiv AI

import os

import requests

 

API_KEY = os.environ.get("IMENTIV_API_KEY")

 

headers = {

    "Authorization": f"Bearer {API_KEY}",

    "Referer: https://imentiv.ai/ "

 

}

 

# Step 1: Upload video — correct endpoint is POST /v2/videos

with open("interview.mp4", "rb") as video_file:

    response = requests.post(

        "https://api.imentiv.ai/v2/videos",

        headers=headers,

        files={"file": ("interview.mp4", video_file, "video/mp4")},

    )

 

result = response.json()

print(result)  # Inspect the raw response to confirm the video_id field name

 

video_id = result["video_id"]  # Verify this field name from the printed response

 

# Step 2: Retrieve per-frame emotion data

frames_response = requests.get(

    f"https://api.imentiv.ai/v1/videos/{video_id}/frames",

    headers=headers,

)

frames_data = frames_response.json()

print(frames_data)  # Inspect structure before parsing

 

# Step 3: Retrieve multimodal analytics (facial + audio + transcript combined)

analytics_response = requests.get(

    f"https://api.imentiv.ai/v2/videos/{video_id}/multimodal-analytics",

    headers=headers,

)

analytics = analytics_response.json()

print(analytics)

 

# Step 4: Retrieve valence-arousal data

valence_arousal_response = requests.get(

    f"https://api.imentiv.ai/v1/videos/{video_id}/valence_arousal",

    headers=headers,

)

valence_arousal_data = valence_arousal_response.json()

print(valence_arousal_data)

  

The core pattern is familiar authenticate, upload media, retrieve structured emotion data. The main difference is that Imentiv AI uses a standard REST pattern, which makes integration more straightforward in most languages. Check the full API reference at  api.imentiv.ai/docs  for exact request/response schemas.

 

Ready to migrate?

Imentiv AI is available now. Sign up, grab your API key, and make your first emotion analysis call in minutes.

Get started at imentiv.ai →