live-enroll¶
Interactive command-line phrase spotter enrollment, using the default audio capture device.
Usage¶
Enrolls TrulyNatural SDK wake words on live audio.
usage: live-enroll -t task [options] +user1 [+user2 ...] [file ...]
options:
-e enrollments : enrollment context output filename
-o out : enrolled model output filename (default: enrolled-sv.snsr)
-p prefix : capture each enrollment to file as
<prefix>-<user>-{pass,fail}-<index>.wav
-s setting=value : override a task setting
-t task : specify task filename (required)
-v [-v [-v]] : increase verbosity
Enrollment audio is captured from the default microphone, unless
the optional [file ...] arguments are supplied.
Settings are strings used as keys to query or change task behavior.
Most frequently used for enrollment is accuracy, which takes a value
between 0 and 1.
Refer to the TrulyNatural SDK documentation for a complete list and
descriptions of all supported settings.
Examples¶
Enroll two phrases interactively on a Raspberry Pi 3.
% cd sample/c
% make -s -j4 all
% bin/live-enroll -v -t ../../model/udt-universal-3.67.1.0.snsr \
+hey-sensory +hello-voice-genie
Say the enrollment phrase (1/4) for "hey-sensory"
Recording: 3.46 s
Preliminary enrollment checks passed.
Say the enrollment phrase (2/4) for "hey-sensory"
Recording: 3.41 s
Preliminary enrollment checks passed.
Say the enrollment phrase (3/4) for "hey-sensory" with context,
for example: "<phrase> will it rain tomorrow?"
Recording: 4.30 s
This enrollment recording is not usable.
Reason: silence-begin
Fix: Please wait for the prompt before speaking.
Say the enrollment phrase (3/4) for "hey-sensory" with context,
for example: "<phrase> will it rain tomorrow?"
Recording: 4.44 s
Preliminary enrollment checks passed.
Say the enrollment phrase (4/4) for "hey-sensory" with context,
for example: "<phrase> will it rain tomorrow?"
Recording: 4.18 s
Preliminary enrollment checks passed.
Say the enrollment phrase (1/4) for "hello-voice-genie"
Recording: 2.30 s
Preliminary enrollment checks passed.
Say the enrollment phrase (2/4) for "hello-voice-genie"
Recording: 3.53 s
Preliminary enrollment checks passed.
Say the enrollment phrase (3/4) for "hello-voice-genie" with context,
for example: "<phrase> will it rain tomorrow?"
Recording: 4.22 s
Preliminary enrollment checks passed.
Say the enrollment phrase (4/4) for "hello-voice-genie" with context,
for example: "<phrase> will it rain tomorrow?"
Recording: 5.59 s
Preliminary enrollment checks passed.
Adapting: 100% complete.
Enrolled model saved to "enrolled-sv.snsr"
Test: