1 adintool
2ADINTOOL(1) General Commands Manual ADINTOOL(1)
4
8 adintool
9 - a tool to record / split / send / receive audio streams
10
12 adintool {-in inputdev} {-out outputdev} [options...]
13
15 adintool analyzes speech input, finds speech segments skipping silence,
16 and records the detected segments in various ways. It performs speech
17 detection based on zerocross number and power (level), and records the
18 detected parts to files or other output devices sucessively.
19 adintool is a upper version of adinrec with various functions.
21 Supported input device are: microphone input, a speech file, standard
22 tty input, and network socket (called adin-net server mode). Julius
23 plugin can be also used. Detected speech segments will be saved to
24 output devices: speech files, standard tty output, and network socket
25 (called adin-net client mode). For example, you can split the incoming
26 speech to segments and send them to Julius to be recognized.
27 Output format is WAV, 16bit (signed short), monoral. If the file
29 already exist, it will be overridden.
30
32 All Julius options can be set. Only audio input related options are
33 treated and others are silently skipped. Below is a list of options.
34 adintool specific options
36 -freq Hz
37 Set sampling rate in Hz. (default: 16,000)
38 -in inputdev
40 Audio input device. "mic" to capture via microphone input, "file"
41 for audio file input, and "stdin" to read raw data from
42 standard-input. For file input, file name prompt will appear after
43 startup. Use "adinnet" to make adintool as "adinnet server",
44 receiving data from client via network socket. Default port number
45 is 5530, which can be altered by option "-inport".
46 Alternatively, input device can be set by "-input" option, in which
48 case you can use plugin input.
49 -out outputdev
51 Audio output device store the data. Specify "file" to save to file,
52 in which the output filename should be given by "-filename". Use
53 "stdout" to standard out. "adinnet" will make adintool to be an
54 adinnet client, sending speech data to a server via tcp/ip socket.
55 "vecnet" will make adintool to be a vecnet client, sending feature
56 vectors extracted from input to a server via tcp/ip socket. When
57 using "adinnet" and "vecnet" output, the server name to send data
58 should be specified by "-server". The default port number is 5530,
59 which can be changed by "-port" option.
60 -inport num
62 When adintool becomes adinnet server to receive data (-in adinnet),
63 set the port number to listen. (default: 5530)
64 -server [host] [,host...]
66 When output to adinnet server (-out adinnet), set the hostname. You
67 can send to multiple hosts by specifying their hostnames as
68 comma-delimited list like "host1,host2,host3".
69 -port [num] [,num...]
71 When adintool send a data to adinnet server (-out adinnet), set the
72 port number to connect. (default: 5530) For multiple servers,
73 specify port numbers for all servers like "5530,5530,5531".
74 -filename file
76 When output to file (-out file), set the output filename. The
77 actual file name will be as "file.0000.wav" , "file.0001.wav" and
78 so on, where the four digit number increases as speech segment
79 detected. The initial number will be set to 0 by default, which can
80 be changed by "-startid" option. When using "-oneshot" option to
81 save only the first segment, the input will be saved as "file".
82 -startid number
84 At file output, set the initial file number. (default: 0)
85 -oneshot
87 Exit after the end of first speech segment.
88 -nosegment
90 Do not perform speech detection for input, just treat all the input
91 as a single valid segment.
92 -raw
94 Output as RAW file (no header).
95 -autopause
97 When output to adinnet server, adintool enter pause state at every
98 end of speech segment. It will restart when the destination adinnet
99 server sends it a resume signal.
100 -loosesync
102 When output to multiple adinnet server, not to do strict
103 synchronization for restart. By default, when adintool has entered
104 pause state, it will not restart until resume commands are received
105 from all servers. This option will allow restart at least one
106 restart command has arrived.
107 -rewind msec
109 When input is a live microphone device, and there has been some
110 continuing input at the moment adintool resumes, it start recording
111 backtracking by the specified milliseconds.
112 -paramtype parameter_type
114 When output is a vecnet (-out vecnet), specify parameter type in
115 HTK format like "MFCC_E_D_N_Z".
116 -veclen vector_length
118 When output is a vecnet (-out vecnet), specify vector length
119 (dim.).
120 Concerning Julius options
122 -input {mic|rawfile|adinnet|stdin|netaudio|esd|alsa|oss}
123 Choose speech input source. Specify 'file' or 'rawfile' for
124 waveform file. On file input, users will be prompted to enter the
125 file name from stdin.
126 'mic' is to get audio input from a default live microphone device,
128 and 'adinnet' means receiving waveform data via tcpip network from
129 an adinnet client. 'netaudio' is from DatLink/NetAudio input, and
130 'stdin' means data input from standard input.
131 At Linux, you can choose API at run time by specifying alsa, oss
133 and esd.
134 -lv thres
136 Level threshold for speech input detection. Values should be in
137 range from 0 to 32767. (default: 2000)
138 -zc thres
140 Zero crossing threshold per second. Only input that goes over the
141 level threshold (-lv) will be counted. (default: 60)
142 -headmargin msec
144 Silence margin at the start of speech segment in milliseconds.
145 (default: 300)
146 -tailmargin msec
148 Silence margin at the end of speech segment in milliseconds.
149 (default: 400)
150 -zmean
152 This option enables DC offset removal.
153 -smpFreq Hz
155 Set sampling rate in Hz. (default: 16,000)
156 -48
158 Record input with 48kHz sampling, and down-sample it to 16kHz
159 on-the-fly. This option is valid for 16kHz model only. The
160 down-sampling routine was ported from sptk. (Rev. 4.0)
161 -NA devicename
163 Host name for DatLink server input (-input netaudio).
164 -adport port_number
166 With -input adinnet, specify adinnet port number to listen.
167 (default: 5530)
168 -nostrip
170 Julius by default removes successive zero samples in input speech
171 data. This option stop it.
172 -C jconffile
174 Load a jconf file at here. The content of the jconffile will be
175 expanded at this point.
176 -plugindir dirlist
178 Specify which directories to load plugin. If several direcotries
179 exist, specify them by colon-separated list.
180
182 ALSADEV
183 (using mic input with alsa device) specify a capture device name.
184 If not specified, "default" will be used.
185 AUDIODEV
187 (using mic input with oss device) specify a capture device path. If
188 not specified, "/dev/dsp" will be used.
189 PORTAUDIO_DEV
191 (portaudio V19) specify the name of capture device to use. See the
192 instruction output of log at start up how to specify it.
193 LATENCY_MSEC
195 Try to set input latency of microphone input in milliseconds.
196 Smaller value will shorten latency but sometimes make process
197 unstable. Default value will depend on the running OS.
198
200 Record microphone input to files: "data.0000.wav", "data.0001.wav" and
201 so on:
202 Split a long speech file "foobar.raw" into "foobar.1500.wav",
203 "foobar.1501.wav" ...:
204 Copy an entire audio file via network socket.
205 Detect speech segment, send to Julius via network and recognize it:
206
208 julius ( 1 ) ,
209 adinrec ( 1 )
210
212 Copyright (c) 1991-2013 Kawahara Lab., Kyoto University
213 Copyright (c) 1997-2000 Information-technology Promotion Agency, Japan
215 Copyright (c) 2000-2005 Shikano Lab., Nara Institute of Science and
217 Technology
218 Copyright (c) 2005-2013 Julius project team, Nagoya Institute of
220 Technology
221
223 The same as Julius.
224 12/19/2013 ADINTOOL(1)