| # Assistant Audio Metadata File Format. |
| |
| Each audio metadata file is used to describe the recorded audio samples. These |
| metadata files should be of json format. |
| |
| ## Allowed field names and value formats |
| |
| The following field names and corresponding value formats are allowed in the |
| metadata file: |
| |
| * audio_file_path - A string; This should be the absolute path/uri of the |
| recorded audio sample. |
| * script - A string; This should be the text used to record the audio sample. |
| eg: Who is obama? |
| * number_of_channels - An Integer; The number of channels, the recorded audio |
| sample has. |
| * sample_format - A string; The sample format of recorded stream container. |
| eg: S32LE |
| * sample_rate - An Integer; The sampling rate of recorded audio sample in hertz. |
| |
| ## Sample File |
| |
| { |
| "audio_file_path": "okgooglesetalarmat5am.raw", |
| "script": "set alarm at 5 a.m.", |
| "number_of_channels": "8", |
| "sample_format": "S32LE", |
| "sample_rate": "48000" |
| } |