News: Downloadable PDA Demos
Introduction
Technical Description
How To
Documents
Product Brief
White Paper
Supported Platforms
Consulting
License Terms
NFL Demo
Downloads
Secure Travel Demo
FAQ's for Users
&
Developers
Registration Form and License Agreement
SDK Access
Client Access
Feedback
Other Sprex Products/Services
Exit
|
SPRECC(1) Sprecc 1.3: Man Page Version 5 SPRECC(1)
NAME
sprecc - SPeech RECognition Client
SYNOPSIS
sprecc -i ip_address [ -p port ] [ -f input_file ] [ -d
device_path ] [ -s seconds ] [ -n network_file ] [ -N net-
work_id ] [ -v verbosity ] [ -ecEMbmrzh? ]
DESCRIPTION
sprecc is a client for sprecd. It opens a specified file
or live audio device, sending its audio data over a TCP
socket to a specified IP address and port number and
prints the returned result on the standard output.
OPTIONS
-i ip_address [required] Specify the ip address in dot-
ted decimal form, as in 200.100.50.25, or the hostname, as
in: cassandra.sprex.com
-p port [optional] Specify the port number. If not
specified, getservbyname() is called with /etc/services
for an entry like there, then the default value of 4101 is
used.
-f input_file [optional] Read the data from this file.
The data should be headerless 16-bit PCM or 8-bit mulaw at
8000 samples per second, according to what the server
requires.
-d device_path [optional] Read the data from this
device file. Interrupt using control-C. May not be
available under Windows (optional); run "sprecc -h" to
check.
-s seconds [optional] If reading from -d device_path
then stop reading after seconds seconds, as an alternative
to control-C.
-n network_file [optional] Read an HTK SLF file (see
the HTK Book) from this file and send it to the server to
use for recognizing the audio.
-N network_id [optional] Tell the server to get the
network to use from the given network_id, whether in uni-
versal form or as a local filename path relative to the
server's grammar root directory. The network_id is a
Grammar Resource Locator (GRL).
-v verbosity Verbosity of debugging output (0=low,
1=medium or 2=high) [default: 0].
-e [optional] Use HAPI endpointer to endpoint
speech locally.
-c [optional] Calibrate HAPI endpointer
before recording speech to send to recognizer. Calibra-
tion is used for endpointing so this implies -e; if -e was
not specified, it will be set.
-E [optional] Send EPC parameters instead of
raw audio.
-M [optional] Send MFCC parameters instead of
raw audio.
-b [optional] Byteswap PCM data because the
server is expecting x86 byte order.
-m [optional] If reading audio data from a
device then configure that device to produce 8000Hz mulaw
data instead of the default 16000Hz 16-bit PCM data. May
not be available under Windows (optional); run "sprecc -h"
to check.
-r [optional] Start recording before socket
connect. May not be available under Windows (optional);
run "sprecc -h" to check.
-z [optional] Compress MFCC or EPC parameters
sent. Not available unless -M or -E options selected.
May not be available under Windows (optional); run "sprecc
-h" to check.
-h [optional] Show help information and exit.
(Same effect is produced if you give no command-line argu-
ments to sprecc.)
-? [optional] Show help information and
exit. (Same effect is produced if you give no command-
line arguments to sprecc.)
EXAMPLES
To get a list of options and their explanation, use no
options:
example% sprecc
To send sample.sw to sprecd running on 127.0.0.1 (local-
host) at port 4102, do:
example% sprecc -i 127.0.0.1 -p 4102 -f sample.sw
To recognize using live audio using sprecd at
server1.sprex.com, port 4099 do:
example% sprecc -i server1.sprex.com -p 4099 -d
/dev/audio
Then speak. Then type control-C one time to end audio
input. Or twice to kill the process before completion.
BUGS
None known. Please report bugs to tv@sprex.com Your feed-
back on this document is solicited.
AUTHOR
Tom Veatch, Sprex, Inc. All rights reserved. sprecc is
licensed software, to be used only under written license
from Sprex.
Product of Sprex, Inc. 26 Oct. 2003 SPRECC(1)