Celebrity Voice Detection

Overview

Our Celebrity Voice Detection API detects the presence of well-known figures in audios and videos. Powered by our voice recognition technology, Celebrity Voice Detection identifies over 500 celebrities, politicians, and well-known public figures.

Given an input audio or video file, the model analyzes the voice and identifies potential matches to individuals in our index. For each detected individual, the API returns the matching name and confidence score. It also returns an attribute category, which describes what type of famous figure the recognized individual is. The available categories are: film, music, media, politics, business, and sports.

Celebrity Voice Detection can help identify when audio contains or recreates the recognizable voice of a public figure, including in generated or manipulated content.

Request Format

# submit a task with media with url
curl --location --request POST "https://api.thehive.ai/api/v2/task/sync" \
  --header "accept: application/json" \
  --header "authorization: token <YOUR_API_KEY>" \
  --form "url="http://hive-public.s3.amazonaws.com/test/speech_test.mp3"'

# submit a task with media with local media file
curl --location --request POST "https://api.thehive.ai/api/v2/task/sync" \
  --header "accept: application/json" \
  --header "authorization: token <YOUR_API_KEY>" \
  --form 'media=@"<absolute/path/to/file>"'

Response

The Celebrity Voice Detection API returns recognition results for each sampled timestamp in the media. For each timestamp, the response includes the detected individual's name, confidence score, and category. To see an annotated example of an API response object for this model, visit our API reference page.

Thresholds

We recommend the following threshold for optimized model performance: 0.875.
We suggest starting with this threshold, but you should always refine these thresholds to suit your specific use case.