This section explains how to make a request using a file. For a URL, the process is similar, with the main difference being the endpoint to call in step 1.
Submission guidelines
Given the highly variable nature of audio, it is advisable to follow these simple guidelines:- Send to the system audio containing only speech, as the analysis models are currently trained to recognize speech and not other types of audio (such as music or noise);
- For greater accuracy, it is suggested to send audio of an appropriate duration to the system, such as at least 10 seconds and not less than 3 seconds;
- The most accurate results are obtained by sending clean audio to the system, thus avoiding background noise (traffic noise or excessively loud noise, which could compromise the analysis).
Submit a speech using API
To classify a speech, you need to make a POST request to/api/classification_audio with the required parameters (for details on the parameters, refer to the API Reference section).
Analyze specific moments
By default the speech is split into chunks analyzed over its whole duration. If only some moments are relevant, list them inaudio_timestamps: the system analyzes a short audio chunk centred on each of them.
- Timecodes are comma-separated and in
HH:MM:SS.mmmformat (hours, minutes, seconds and milliseconds), measured from the beginning of the audio. Duplicates are counted once. - Timecodes close enough to fall within the same chunk are analyzed together in a single chunk.
- One audio chunk is charged for each distinct timecode; chunks that are not produced (grouped timecodes, timecodes beyond the end of the audio) are refunded once the analysis is complete. If none of the timecodes falls within the audio, the classification ends with an error.
- You can list up to 30 timecodes if your pricing plan lets you choose the number of audio chunks, otherwise up to 5.
Specific moments are a pricing plan feature: if your plan does not include them,
audio_timestamps is ignored. Malformed timecodes or more timecodes than allowed are rejected with a 400 error.Retrieve the results
Use the provided identifier to retrieve the classification results by making a GET request to/api/classification_audio/{identifier} (for details on how to structure this request, see the API Reference section).
The classification may not be finished yet! If the classification is not yet complete, continue sending the GET request until the result is available.
Guidelines on interpreting the results
As for interpreting the results, the advice is to focus on:- Phantom as the primary result for the verdict (platform’s default choice);
- Zebra II as the secondary result for the verdict;
See also
Classify an image
Find out how to start an image classification and discover whether your image is authentic or artificially generated
Classify a video
Find out how to start a video classification and discover whether your video is authentic or artificially generated