Call log directories and files
For each session, Speech Server creates a directory for all log files pertaining to that session, including:
-
Call log text files
- Grammars
- Acoustic models
- Captured waveforms (audio files)
- Diagnostic files
Disk space requirements
The system manages disk space on the entire disk or partition on the Speech Server host. When the maximum space is approached, Speech Server issues a warning, and when the limit is reached, Speech Server stops recording call log files, utterances, and whole call recordings (if enabled). To configure the thresholds by adding server.callLog.minFreeDiskMB and server.callLog.warnMinFreeDiskMB.
The system records utterances on Speech Server hosts, and saves them in the same location as the call log file for the current call. The Speech Server saves 100% of all caller utterances by default. One way to limit the amount of disk space occupied by utterances is to set server.callLog.utteranceSampling.
Log directory paths
For each session, the system creates a path to store all call log files for that session. It writes call log data, utterances, and whole call recordings (if enabled) to a hierarchical directory structure on Speech Server hosts.
The path has this format:
base/CompanyName/callLogs/ApplicationName/yyyy/mmMonth/dd/hh/
|
Token |
Description |
|---|---|
|
base |
The root path for storing files as set by server.callLog.baseDirectory in NSSserver.cfg. Default: $NUANCE_DATA_DIR/ |
|
CompanyName |
The company name as set by company_name in session.xml. See Configuring sessions (session.xml). If not set here, the default is Nuance. |
|
ApplicationName |
The application name as set by application_name in session.xml. See Configuring sessions (session.xml). If not set here, the default is MyApp. |
|
yyyy/mm/dd/hh |
The year, month, day, and hour when the session began. The month directory format is ddMonth. For example, 09September. The hour directory format uses a 24-hour clock. All the calls that occur within the same hour of a date are written to the same directory. For example, calls that started between 4 and 5 p.m. on May 10, 2012 are written to . . .\2012\05May\10\16\ |
The call log directory contains various types of files.
Note: The system encodes call logs as UTF-8. Treat these files as binary for storage and transport purposes (for example, storing in CVS and transporting via ftp).
Log file naming
Call log filenames indicate minutes and seconds, the host, and a unique session ID generated at the start of the session.
Logged utterances are saved in a RIFF format in the same directories as their call log files on Speech Server hosts.
|
File type |
Filename format |
|---|---|
|
Call log text file |
NUAN-mm-ss-host-sid-appsessionid-LOG |
|
Encrypted call log text file |
NUAN-mm-ss-host-sid-appsessionid-LOG.enc |
|
Post-endpointing utterance |
NUAN-mm-ss-host-sid-appsessionid-uttXXX-POSTEP.wav (Nuance Recognizer) |
|
Post-endpointing utterance |
NUAN-mm-ss-host-sid-appsessionid-uttXXXnlp-POSTEP.wav (Krypton, via NLP service) |
|
Pre-endpointing utterance |
NUAN-mm-ss-host-sid-appsessionid-uttXXX-PREEP.wav |
|
Endpointing-info utterance |
NUAN-mm-ss-host-sid-appsessionid-uttXXX-EPINFO.txt |
|
Save-Waveform |
NUAN-mm-ss-host-sid-appsessionid-uttXXX-SAVEWAVEFORM.wav |
|
ExposePlayable |
NUAN-mm-ss-host-sid-appsessionid-uttXXX-EXPOSEPLAYABLE.wav |
|
ExposeRecording |
NUAN-mm-ss-host-sid-appsessionid-uttXXX-EXPOSERECORDING.wav |
|
Recording |
NUAN-mm-ss-host-sid-appsessionid-uttXXX-RECORDING.wav |
|
Dictations |
NUAN-mm-ss-host-sid-appsessionid-uttXXX-DICTATION.wav |
|
Whole Call Recording |
NUAN-mm-ss-host-sid-appsessionid-WCR.wav |
- Encrypted files have the suffix
.enc. - Whole call recording files may have different suffixes in the different WCR modes.
Filenames use the following tokens:
|
Token |
Description |
|
mm |
Number of minutes into the current hour. |
|
ss |
Number of seconds into the current minute. |
|
host |
Name of the host writing the log record. |
|
sid |
Unique session identifier created by the system. |
|
appsessionid |
Unique session identifier created by the application, if one is set. See Nuance-specific capabilities. |
|
uttXXX |
Waveforms are written in .wav format using the naming scheme utt001.wav, utt002.wav, and so on. (If the count exceeds 999, it rolls over to 000, 001, and so on.) |
