Handler transcription settings

Each handler configuration has a separate section that specifies a set of parameters to interact with V-Blaze. The default [Adapter] section within /config/default.cfg contains two handlers to separate transcription requests:

[LiveCall] — Activates at the start of a call. The [LiveCall] handler contains transcription settings optimized for speed.
[PostCall] — Activates when a call has completed. The [PostCall] handler contains transcription settings optimized for quality.
[LivePostCall] — Activates at the start of a call. The [LivePostCall] handler streams audio to V-Blaze for transcription while the call is in progress. The call audio does not write to disk, which means that confidential information is never stored. Post call transcripts are quickly available because the transcription process begins at the start of the call.

The following is a sample [LiveCall] handler:

[LiveCall]
server = localhost
actions = LiveLog
activitylevel = 50
uttminactivity = 250
uttmaxsilence = 500
uttpadding = 250
uttmaxtime = 15
uttmaxgap = 0
scrubtext = true
model = eng-us:callcenter
protocol = wss
streaming = true

The following is a sample [PostCall] handler:

[PostCall]
server = localhost
actions = PostLog
gender = true
emotion = true
scrubdir = /var/lib/vociadapter/json
utthandler = true
mincalldur = 10
maxcalldur = 7200
maxcalls = 100000

The following is a sample


[LivePostCall]

handler:

[LivePostCall]
server = localhost
actions = PostLog
livepost = true
protocol = ws
streaming = true
scrubdir= /tmp/audio
scrubtext = true
scrubconf = /opt/voci/adapter/config/scrub.conf

Table 1. Handler Settings — /config/custom.cfg
Parameter	Values	Description
server	localhost (default), URL: HTTP or HTTPS	Specify the hostname or address of the ASR engine for the handler to use. Port numbers are supported.
actions	No default	Specify a list of actions for the handler to use. Actions contain a set of parameters that define what to do with transcripts.
checkinstall	true, false	Validates the `actions` specified in the handler configuration when set to `true`.
billing	string	Specify a name that indicates who the handler is configured for.
billing_agentid	string	Specify an OrecX variable by enclosing the name of the variable within curly brackets. To use the OrecX `agent_id` value to populate the field, set the value to `{agent_id}`.
model	eng-us:callcenter (default)	Specify the preferred language model to use.
scrubtext	true, false	V-Blaze API parameter — Enables the ASR engine to redact sensitive information from transcripts.
scrubdir	/tmp/audio (default)	Specify a directory to save redacted transcripts.
utthandler	true (default), false	If `utthandler=true` (default), actions designed to work with the live-stream URL run whenever an utterance is transcribed. If `utthandler=false`, actions run when the call completes and the full transcript is available.
mincalldur	10 (default)	Specifies the minimum call length (seconds) for the handler to accept. `mincalldur` doesn't apply to real-time configurations.
maxcalldur	7200 (default	Specifies the maximum call length (seconds) for the handler to accept.
maxcalls	100000	Specifies the maximum number of simultaneous calls the handler allows.
filters	No default	Specifies which filter configurations to call.
gender	true, false	V-Blaze API parameter — Set `gender=true` to enables gender detection. For each utterance, the ASR engine will attempt to determine if the speaker is male or female.
emotion	true, false	V-Blaze API parameter — Set `emotion=true` to enable emotion detection. Each utterance will be categorized as strongly positive, positive, neutral, negative, or strongly negative.
livepost	true, false	Enables audio streaming at the start of a call. When enabled, call audio does not save to storage. `livepost` is intended to be used in conjunction with `streaming`.
streaming	true, false	When set to `false`, the adapter sends a live stream URL to V-Blaze, and V-Blaze connects to the stream to transcribe it. When set to `true`, the adapter captures the live stream and forwards it to V-Blaze to transcribe. `streaming` requires the `protocol` option to be enabled.
protocol	ws, wss	Specify the WebSocket protocol to use with the streaming option.
reconnectdelay	1 (default)	Initial delay (seconds) for reconnect attempts.
reconnectattempts	0 (default)	Maximum number of reconnect attempts before shutting down the service. For unlimited attempts, set `reconnectattempts=0`.
reconnectmaxdelay	600 (default)	Upperbound in seconds for the exponential backoff for reconnect attempts.

Note: Refer to V-Blaze Transcription Parameters for more information on the V-Blaze API parameters in the table above.