Skip to main content
POST
/
predict
AI-voice detection
curl --request POST \
  --url https://api.aurigin.ai/v0/predict \
  --header 'Content-Type: multipart/form-data' \
  --header 'x-api-key: <api-key>' \
  --form file='@example-file' \
  --form user_id=speaker_123
{
  "predictions": [
    "fake",
    "fake",
    "real"
  ],
  "global_probability": [
    0.9584,
    0.9585,
    0.9123
  ],
  "error": [
    null,
    null,
    null
  ],
  "model": "apollo-4-2025-10-20",
  "processing_time": 1.350719928741455,
  "audio_duration": 69.91
}

Documentation Index

Fetch the complete documentation index at: https://docs.aurigin.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

The /predict endpoint analyzes audio files for AI-generated content (deepfakes, voice cloning, text-to-speech). Returns results in real-time for small files.
File Size Limits: For very large files (100MB+), use pre-signed S3 URLs for asynchronous processing.

​
Authentication

​
X-Api-Key
string
required
Your API key for authentication

​
Request Parameters

​
file
file
required
Audio or video file to analyzeSupported formats: WAV, MP3, AAC, FLAC, OGG, M4A, MP4, MOV, AVI, MKVMax size: 5 MB (recommended for fastest processing)
​
device
string
required
Device type making the requestOptions: macos, windows, web_app, api
​
prediction_id
string
Custom prediction ID for tracking (auto-generated if not provided)Format: pred_ followed by 12 hex charactersExample: pred_9b6ff057a7f7
​
model
string
default:"stable"
Model version to useOptions:
  • stable - Production model (recommended)
  • stable-latest - Latest stable release
  • dev-v4 - Development model (testing only)

​
Response

​
prediction_id
string
required
Unique identifier for this prediction
​
global
object
required
Overall prediction for the entire audio file
​
segments
array
required
Per-segment analysis (one per chunk_duration)
​
model
string
required
Model version used for prediction
​
processing_time
number
required
Time taken to process in seconds
​
audio_duration
number
required
Total audio file duration in seconds
​
warnings
array
List of warnings (e.g., partial chunk failures, quality issues)

​
Example Request

curl -X POST "https://api.aurigin.ai/v1/predict" \
  -H "X-Api-Key: YOUR_API_KEY" \
  -F "file=@suspicious_voice.wav" \
  -F "device=api" \
  -F "model=stable"

​
Example Response

200 - Spoofed Audio Detected
{
  "prediction_id": "pred_9b6ff057a7f7",
  "global": {
    "confidence": 0.95,
    "result": "spoofed",
    "reason": null
  },
  "segments": [
    {
      "index": 0,
      "start": 0.0,
      "end": 5.0,
      "confidence": 0.96,
      "result": "spoofed"
    },
    {
      "index": 1,
      "start": 5.0,
      "end": 10.0,
      "confidence": 0.94,
      "result": "spoofed"
    }
  ],
  "model": "stable-latest",
  "processing_time": 1.23,
  "audio_duration": 10.0,
  "warnings": []
}
200 - Authentic Audio
{
  "prediction_id": "pred_abc123def456",
  "global": {
    "confidence": 0.92,
    "result": "bonafide",
    "reason": null
  },
  "segments": [
    {
      "index": 0,
      "start": 0.0,
      "end": 5.0,
      "confidence": 0.91,
      "result": "bonafide"
    },
    {
      "index": 1,
      "start": 5.0,
      "end": 8.5,
      "confidence": 0.93,
      "result": "bonafide"
    }
  ],
  "model": "stable-latest",
  "processing_time": 0.87,
  "audio_duration": 8.5,
  "warnings": []
}
400 - Validation Error
{
  "error": "validation_error",
  "message": "Unsupported audio format",
  "correlation_id": "a1b2c3d4-e5f6-7890-abcd-ef1234567890"
}
401 - Unauthorized
{
  "error": "unauthorized",
  "message": "Invalid or inactive API key",
  "correlation_id": "8b738992-c4eb-4f19-870a-900e6830d147"
}
413 - File Too Large
{
  "error": "payload_too_large",
  "message": "Request body exceeds maximum allowed size of 5MB",
  "correlation_id": "7f629881-a3db-5e28-761b-811f7940c258"
}

​
Error Codes

CodeDescriptionSolution
400Validation errorCheck file format, duration, parameters
401UnauthorizedVerify API key is valid and active
413File too largeReduce file size or use pre-signed S3 URLs
422Unsupported formatConvert to supported audio format
500Processing failedRetry or contact support if persists
503Service unavailableRetry with exponential backoff

​
Confidence Score Interpretation

The confidence score indicates how certain the model is about its prediction:
ConfidenceInterpretationAction
0.9 - 1.0Very confidentTrust the result
0.7 - 0.9ConfidentGenerally reliable
0.4 - 0.7ModerateReview segments individually
0.0 - 0.4Low confidenceManual review recommended
For high-security applications, set a threshold of 0.85+ before taking automated actions.

​
Best Practices

  • Use WAV or FLAC for best accuracy
  • Keep files under 5MB for fastest processing with this endpoint
Check the warnings array for partial failures:
"warnings": ["3 of 10 segments skipped due to quality issues"]
If warnings exist, review individual segments for confidence.
Implement exponential backoff for transient errors:
import time
from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry

session = requests.Session()
retry = Retry(
    total=3,
    backoff_factor=1,
    status_forcelist=[500, 502, 503, 504]
)
adapter = HTTPAdapter(max_retries=retry)
session.mount('https://', adapter)
Track processing_time to identify slow requests:
  • < 2s: Excellent
  • 2-5s: Good
  • 5-10s: Acceptable for large files
  • > 10s: Consider async processing

​
Rate Limits

PlanRequests/MinuteConcurrent
Free102
Starter605
Professional30020
EnterpriseCustomCustom
Upgrade your plan at app.aurigin.ai/billing for higher limits.

Pre-Signed URLs

Asynchronous processing for large files (100MB+)

Authorizations

​
x-api-key
string
header
required

Body

​
file
file
required
​
user_id
string

Optional user identifier

Response

OK

​
error
(string | null)[]

Error messages for each 5-second chunk (null if successful). Aligns 1:1 with the predictions array.

​
global_probability
number<float>[]

Confidence scores (0.0-1.0) for each prediction, one per 5-second chunk. Aligns 1:1 with the predictions array.

​
predictions
enum<string>[]

AI detection results for each 5-second chunk of the audio. Array length equals the number of 5-second chunks in the audio file.

Available options:
fake,
real