Audio Segment
Audio segments attempts to subdivide a song into many segments, with each segment containing a roughly consistent sound throughout its duration.
Parameters
The starting point (in seconds) of the segment.
The duration (in seconds) of the segment.
The confidence, from 0.0 to 1.0, of the reliability of the segmentation. Segments of the song which are difficult to logically segment (e.g: noise) may correspond to low values in this field.
The onset loudness of the segment in decibels (dB). Combined with loudness_max and loudness_max_time, these components can be used to desctibe the “attack” of the segment.
The segment-relative offset of the segment peak loudness in seconds. Combined with loudness_start and loudness_max, these components can be used to desctibe the “attack” of the segment.
The peak loudness of the segment in decibels (dB). Combined with loudness_start and loudness_max_time, these components can be used to desctibe the “attack” of the segment.
The offset loudness of the segment in decibels (dB). This value should be equivalent to the loudness_start of the following segment.
A “chroma” vector representing the pitch content of the segment, corresponding to the 12 pitch classes C, C#, D to B, with values ranging from 0 to 1 that describe the relative dominance of every pitch in the chromatic scale
Timbre is the quality of a musical note or sound that distinguishes different types of musical instruments, or voices. Timbre vectors are best used in comparison with each other.