The general syntax for naming a data stream instance is:
StreamX
where:
Stream
|
is the prefix for the type of data stream. |
| X | is the instance number. The instance number cannot begin with the digit 0. The number is optional–an instance without an assigned number is still treated as unique and different from other instances that have numbers. |
For example, w1 represents the instance ‘1’ of the word label stream. f3 represents the instance ‘3’ of the speech frame stream.
This table contains a complete list of the available data stream types.
| Data stream type | Prefix |
|---|---|
| Audio sample–mono | a |
| Audio sample–left channel | l |
| Audio sample–right channel | r |
| Feature frame–unnormalized feature | f |
| Feature frame–normalized feature | nf |
| Word tag (for example, recognition result) | w |
| Language identification feature | lf |
| Language identification result | lid |
| Language identification boundary point information | lb |
| Speaker identification output results | sid
|
| Audio fingerprint feature data | fp |
| Audio fingerprint identification data (identified track records) | tr
|
| Phone data for phrase matching | p |
| Score data for phrase matching | s |
| File label boundaries | fl |
| Time shift data | ts
|
|
|