UniMRCP Documentation

UniMRCP is an open source project that provides compliance with the following IETF specifications:

• MRCPv2: RFC6787
• MRCPv1: RFC4463

Scenarios are defined in XML using the <umcscenario> tag. Each scenario must specify a name (used for execution) and a class (which determines the behavior/type of the scenario).

Supported classes include:

• Synthesizer: For text-to-speech tasks.
• Recognizer: For speech recognition tasks.
• Recorder: For recording audio.
• Verifier: For verification tasks.
• DtmfRecognizer: For DTMF (dual-tone multi-frequency) recognition.
• Params: For parameter-related tasks.

<umcscenario name="synth" class="Synthesizer">
<umcscenario name="recog" class="Recognizer">
<umcscenario name="rec" class="Recorder">
<umcscenario name="verify" class="Verifier">
<umcscenario name="dtmf" class="DtmfRecognizer">
<umcscenario name="params" class="Params">

UniMRCP provides two types of documentation for developers:

1. API Reference: Doxygen-generated documentation is located in the docs/dox directory.
2. Design Concepts: UML-based design documentation (HTML pages generated by Enterprise Architect) is located in the docs/ea directory.

The unimrcpclient provides a way to test specific UniMRCP capabilities through its interactive command line. After starting the client, use the run command followed by the application type and an optional profile name.

Supported Application Names:

• synth: Speech Synthesis
• recog: Speech Recognition
• bypass: Bypass mode
• discover: Resource discovery

Usage Pattern: run <app_name> <profile_name>

Note: If profile_name is omitted, it defaults to uni2.

# Run recognition with the 'uni1' profile
>run recog uni1

The UniMRCP Client (UMC) sample application loads scenario files from the conf/umc-scenarios/ directory. To execute a specific scenario, use the run command followed by the unique name attribute assigned to that scenario in its XML definition. Note that while multiple scenarios can share the same class, they must have unique name values to be selectable.

> run synth
> run recog

The asrclient is a command-line tool used to run demo ASR (Automatic Speech Recognition) sessions. You can launch the application with various configuration options for the project root directory, logging priority, and logging output mode.

Command Syntax:

asrclient [options]

Options:

• -r, --root-dir <path>: Set the project root directory path.
• -l, --log-prio <priority>: Set the log priority (0-emergency through 7-debug).
• -o, --log-output <mode>: Set the log output mode:
  • 0: none
  • 1: console only
  • 2: file only
  • 3: both
• -h, --help: Show the help message.

asrclient -r /path/to/root -l 7 -o 1

Logging for the unimrcpclient can be configured via command-line arguments at startup or via interactive commands during a session.

Startup Configuration

Use the following flags to set logging behavior immediately:

• -l [priority]: Sets the log priority (0-7).
• -o [mode]: Sets the log output mode:
  • 0: none
  • 1: console only
  • 2: file only
  • 3: both

Interactive Configuration

While the client is running, use the loglevel command:

>loglevel 5

unimrcpclient -l 7 -o 1

On Windows platforms, uni_service_run is used to wrap the UniMRCP server as a Windows service.

Signature: apt_bool_t uni_service_run(const char *name, apt_dir_layout_t *dir_layout, apr_pool_t *pool);

The uni_daemon_run function is used to start the UniMRCP server in daemon mode. This is typically called by the main entry point when the -d (Linux) or -w (without command-line) flags are provided.

Signature: apt_bool_t uni_daemon_run(apt_dir_layout_t *dir_layout, apt_bool_t detach, apr_pool_t *pool);

The uni_cmdline_run function starts the UniMRCP server in the foreground, allowing for interactive command-line usage. This is the default mode when no daemon or service flags are specified.

Signature: apt_bool_t uni_cmdline_run(apt_dir_layout_t *dir_layout, apr_pool_t *pool);