ADINTOOL(1)

📦 version 4.2.2, in ubuntu jammy, ubuntu noble, ubuntu resolute, ubuntu stonking

✨ ADINTOOL(1) 🎀 adintool ✨

okay bestie, this is the manual page for adintool 💖 It's a tool to record / split / send / receive audio streams 🦋

NAME

adintool — a tool to record / split / send / receive audio streams 🎀

SYNOPSIS

adintool {-in inputdev} {-out outputdev} [options...]

DESCRIPTION

okay so adintool is literally a speech listener 👂✨ It analyzes speech input, finds the speech segments and skips the silence 🙈, then records each detected segment in a bunch of different ways 💅. The detection is based on zerocross number and power (level) 🌈, and the detected parts get saved to files or other output devices one after another 🫶.

adintool is basically the upgraded version of adinrec 🔥 with way more functions. Supported input devices are: microphone input, a speech file, standard tty input, and network socket (called adin-net server mode) 🎧. The Julius plugin can also be used 🧚. Detected speech segments can be saved to output devices: speech files, standard tty output, and network socket (called adin-net client mode) 💜. For example, you can split the incoming speech into segments and send them to Julius to be recognized 👑.

Output format is WAV, 16bit (signed short), monoral 🎀. If the file already exists, it will be overridden 😭, so be careful bestie 💀.

OPTIONS

All Julius options can be set 🌸. Only the audio input related ones are treated, and the rest get silently skipped 🙈. Below is the list ✨.

adintool specific options

  • -freq Hz — Set the sampling rate in Hz 🦋 (default: 16,000).
  • -in inputdev — Picks your audio input device 🎤. "mic" captures from the microphone, "file" is for audio file input, and "stdin" reads raw data from standard input 💖. For file input, a file name prompt pops up after startup 📝. Use "adinnet" to make adintool an adinnet server that receives data from a client over a network socket 🌐. The default port number is 5530, which you can change with the -inport option 🎀.

Alternatively, the input device can be set with the -input option, in which case you can use a plugin input 🧚.

  • -out outputdev — Where the data gets stored 💾. Specify "file" to save to a file, and the output filename is given by -filename 🌸. Use "stdout" to send it to standard out 📤. "adinnet" makes adintool an adinnet client, sending speech data to a server over a TCP/IP socket 🫶. With "adinnet" output, the server name has to be given with -server 🖥️. The default port number is 5530, which can be changed with the -port option 🦋.
  • -inport num — When adintool is an adinnet server receiving data (-in adinnet), this sets the port number to listen on 👂 (default: 5530).
  • -server [host] [,host...] — When outputting to an adinnet server (-out adinnet), this sets the hostname 🏠. You can send to multiple hosts by listing their hostnames as a comma-delimited list, like "host1,host2,host3" 💜.
  • -port [num] [,num...] — When adintool sends data to an adinnet server (-out adinnet), this sets the port number to connect to 🔌 (default: 5530). For multiple servers, give the port numbers for all of them, like "5530,5530,5531" 🌈.
  • -filename file — When outputting to a file (-out file), this sets the output filename 📁. The actual file names will be "file.0000.wav", "file.0001.wav" and so on 🎀, where the four-digit number goes up with each detected speech segment 🔥. The starting number is 0 by default, and you can change it with the -startid option ✨. When you use -oneshot to save only the first segment, the input gets saved as "file" 🦋.
  • -startid number — For file output, sets the initial file number 🍓 (default: 0).
  • -oneshot — Exits after the end of the first speech segment 👋.
  • -nosegment — Skips speech detection entirely and treats all the input as one single valid segment 🫧.
  • -raw — Outputs a RAW file with no header 🧁.
  • -autopause — When outputting to an adinnet server, adintool enters a pause state at the end of every speech segment ⏸️. It restarts when the destination adinnet server sends it a resume signal ▶️.
  • -loosesync — When outputting to multiple adinnet servers, this skips strict synchronization on restart 🤸. By default, once adintool has entered the pause state, it won't restart until resume commands come in from all servers 🙅. This option lets it restart as soon as at least one restart command has arrived 🏃‍♀️.
  • -rewind msec — When the input is a live microphone and there was continuous input at the moment adintool resumes, it starts recording backtracked by the specified number of milliseconds ⏪.

Concerning Julius options

  • -input {mic|rawfile|adinnet|stdin|netaudio|esd|alsa|oss} — Picks the speech input source 🎧. Specify 'file' or 'rawfile' for a waveform file. For file input, you'll be prompted to type the file name from stdin ⌨️.

'mic' grabs audio from a default live microphone device 🎤, and 'adinnet' means receiving waveform data over TCP/IP from an adinnet client 🌐. 'netaudio' is for DatLink/NetAudio input 🔊, and 'stdin' means data comes from standard input 📥.

On Linux, you can choose the API at run time by specifying alsa, oss, and esd 🐧.

  • -chunk_size samples — Audio fragment size in number of samples (default: 1000) 🧩.
  • -lv thres — The level threshold for speech input detection 📊. Values should be in the range from 0 to 32767 (default: 2000).
  • -zc thres — The zero crossing threshold per second 📈. Only input that goes over the level threshold (-lv) gets counted (default: 60).
  • -headmargin msec — Silence margin at the start of a speech segment, in milliseconds 🌷 (default: 300).
  • -tailmargin msec — Silence margin at the end of a speech segment, in milliseconds 🌙 (default: 400).
  • -zmean — Turns on DC offset removal ✨.
  • -smpFreq Hz — Sets the sampling rate in Hz 🎶 (default: 16,000).
  • -48 — Records input at 48kHz sampling and down-samples it to 16kHz on the fly 🎚️. This option is only valid for the 16kHz model 🙅. The down-sampling routine was ported from sptk (Rev. 4.0) 💅.
  • -NA devicename — The host name for DatLink server input (-input netaudio) 🖧.
  • -adport port_number — With -input adinnet, sets the adinnet port number to listen on 👂 (default: 5530).
  • -nostrip — Julius normally removes successive zero samples from the input speech data 🧹. This option stops it 🛑.
  • -C jconffile — Loads a jconf file right here 📂. The contents of the jconffile get expanded at this point 🌸.
  • -plugindir dirlist — Specifies which directories to load plugins from 🧚. If there are several directories, list them colon-separated 🌈.

ENVIRONMENT VARIABLES

  • ALSADEV — (used with mic input via the alsa device) specifies a capture device name 🎙️. If it isn't set, "default" is used 💁.
  • AUDIODEV — (used with mic input via the oss device) specifies a capture device path 🎚️. If it isn't set, "/dev/dsp" is used 🔈.
  • LATENCY_MSEC — Tries to set the input latency of microphone input, in milliseconds ⏱️. A smaller value shortens latency but can sometimes make things unstable 😬. The default depends on the OS you're running 💻.

EXAMPLES

Record microphone input to files: "data.0000.wav", "data.0001.wav" and
so on:

Split a long speech file "foobar.raw" into "foobar.1500.wav",
"foobar.1501.wav" ...:

Copy an entire audio file via network socket.

Detect speech segment, send to Julius via network and recognize it:

SEE ALSO

Copyright (c) 1997-2000 Information-technology Promotion Agency, Japan 🇯🇵

Copyright (c) 1991-2008 Kawahara Lab., Kyoto University 🎓

Copyright (c) 2000-2005 Shikano Lab., Nara Institute of Science and Technology 🏫

Copyright (c) 2005-2008 Julius project team, Nagoya Institute of Technology 💜

LICENSE

The same as Julius 🫶.

bestie, that's the whole ADINTOOL(1) manual 🌈💖 Source version 10/02/2008 📅, and you're officially a adintool expert now 👑✨

📦 where this page comes from