Gender Recognition by Voice API

Turn voice clips into gender insights with one API.

Send either an audio URL or an audio file in the body to get gender, confidence, and usage info. Built for IVR, contact centers, and voice-led apps.

View API Docs

VFile upload or audio URL

VConfidence scores returned

VShort clips often <2s

VSecure over HTTPS

Quickstart (JSON body)

POST JSON

/v1/voice-gender-recognition/apiPOST

// Headers

apiKey:

YOUR_API_KEY

// JSON Body

{
  "url": "https://example.com/audio.wav" 
OR
"file": "audio_file"
}

// Sample Response

{
  "gender": "female",
  "confidence": 95,
  "remainingRequests": 120,
  "success": true
}

Send either url or file (base64) in the JSON body—not both. For large uploads, use multipart/form-data to POST /v1/voice-gender-recognition/api with a "file" field.

Choose an API version

Compare V2 and V3

The existing V1 endpoint remains available for URL or file-based requests. V2 and V3 are multipart file-upload endpoints with speech detection and explicit unknown results when the audio does not contain enough usable speech.

V2 · Whole-clip analysis

1 API request per call

V2 checks for human speech, then uses the ECAPA-TDNN voice model to classify the full recording. Use it for recordings with one speaker. Send one audio file as multipart/form-data in the file field.

Why choose V2 over V1? V1's legacy model always predicts male or female, even for silence, music, or noise. V2 adds speech detection and returns unknown when it detects no speech instead of guessing. It also uses ECAPA-TDNN, which models voice characteristics over time; backend test-voice comparisons found it more accurate than the V1 model. V2 combines voices in the same clip, so choose V3 for multi-speaker recordings.

Endpoint

POST /v1/voice-gender-recognition/api/v2

Successful responses return gender as male, female, or unknown, confidence as a percentage, and remainingRequests. When gender is unknown, the response also includes a message explaining why.

An unknown result is returned when no human speech is detected. It is a successful response, not an API error.

cURL request

curl -X POST 'https://api.genderrecognition.com/v1/voice-gender-recognition/api/v2' \
  -H 'apiKey: YOUR_API_KEY' \
  -F 'file=@/path/to/recording.wav'

Example detected-gender response

{
  "success": true,
  "gender": "female",
  "confidence": 92,
  "remainingRequests": 119
}

Example unknown response

{
  "success": true,
  "gender": "unknown",
  "confidence": 0,
  "message": "No clear speech was detected in the audio, so the gender could not be determined.",
  "remainingRequests": 119
}

V3 · Speaker-aware analysis

2 API requests per call

V3 checks for human speech, identifies speakers, then classifies one selected speaker. It is designed for calls, interviews, and other multi-speaker recordings. The first speaker is selected unless they speak for under 2 seconds, in which case the speaker with the most speaking time is selected. Send one audio file as multipart/form-data in the file field.

Why choose V3 over V1 and V2? Like V2, it avoids gender guesses when speech is missing and uses ECAPA-TDNN instead of the legacy V1 model. Unlike V2, which combines all voices into one prediction, V3 uses speaker diarization to isolate one speaker before classification, reducing interference from other speakers. V3 takes longer and makes a paid Deepgram call for recordings containing speech.

Endpoint

POST /v1/voice-gender-recognition/api/v3

Successful responses return gender as male, female, or unknown, confidence as a percentage, and remainingRequests. When gender is unknown, the response also includes a message explaining why.

An unknown result is returned when no speech is detected or the selected speaker has under 1 second of speech. It is a successful response, not an API error.

cURL request

curl -X POST 'https://api.genderrecognition.com/v1/voice-gender-recognition/api/v3' \
  -H 'apiKey: YOUR_API_KEY' \
  -F 'file=@/path/to/recording.wav'

Example detected-gender response

{
  "success": true,
  "gender": "female",
  "confidence": 92,
  "remainingRequests": 118
}

Example unknown response

{
  "success": true,
  "gender": "unknown",
  "confidence": 0,
  "message": "Not enough clear speech from a single speaker was detected in the audio, so the gender could not be determined.",
  "remainingRequests": 118
}

Voice-native enrichment

Drop voice clips in and enrich records with gender signals, confidence, and remaining usage.

Flexible ingest

Send URLs or inline files in the body, and upload with multipart/form-data when preferred.

Moderation & trust

Flag anomalies or mismatches between declared and detected gender in automated queues.

Built for scale

Handle spikes gracefully with elastic processing and transparent usage limits.

Clean JSON responses

Simple fields: gender, confidence, remainingRequests, success.

Format coverage

Supports common codecs and containers, from wav and mp3 to ogg, m4a, and more.

How it works

Voice gender in three steps.

Provide audio

POST to /api with either a url field or a file field (base64); multipart/form-data is supported for file uploads too.

Get JSON response

Receive gender, confidence, and remainingRequests in seconds.

Scale when ready

Automate ingestion for batch jobs or inline IVR flows with the same responses.

cURL

URL or file in body

curl -X POST 'https://api.genderrecognition.com/v1/voice-gender-recognition/api' \
  -H 'Content-Type: application/json' \
  -H 'apiKey: YOUR_API_KEY' \
  -d '{"url": "https://example.com/audio.wav" 
OR
"file": "audio_file"}'

Prefer sending a file inline? Replace url with "file":"BASE64_AUDIO_BYTES".

JavaScript

JSON body

const apiKey = "YOUR_API_KEY";
const payload = {
  "url": "https://example.com/audio.wav" 
OR
"file": "audio_file"
};

const response = await fetch(
  "https://api.genderrecognition.com/v1/voice-gender-recognition/api",
  {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      apiKey,
    },
    body: JSON.stringify(payload),
  }
);
const result = await response.json();