Data format
What to send SymNexusVoice
The inputs the platform expects, and the labels it learns from.
Input
Input format
Audio files (WAV, MP3, FLAC, OGG, M4A), any sample rate; mono or stereo.
| file | duration | source |
|---|---|---|
| call_20240611_0931.wav | 02:14 | contact center |
| press_line3.flac | 00:10 | factory mic |
file,duration,source
call_20240611_0931.wav,02:14,"contact center"
press_line3.flac,00:10,"factory mic"For training
Label format
For tone or sound types: a label per clip (e.g. calm / agitated). For voice verification: a few reference recordings per enrolled person.
| file | label |
|---|---|
| call_0931_seg3.wav | agitated |
| call_0932_seg1.wav | calm |
Tips for good results
- Record labelled clips in the same conditions as live audio (phone line, microphone).
- Enroll at least three clean recordings per person.
- Get consent for recording and voice verification where required.