Requirements

This topic describes software requirements for Nuance Vocalizer.

This release is known to have these issues and limitations:

  • Nuance tests and supports this product on hosts set to the en-US locale. Running on other locales might affect recognition accuracy and application performance.
  • The limit for running multiple voices simultaneously on a dedicated voice host is 50.

Hardware requirements

The CPU must support the single instruction, multiple data (SIMD) instruction set extension with streaming SIMD Extensions (SSE v4.2).

Licensing requirements

Vocalizer requires a License Manager and license files to deliver them. You must acquire licenses and run license servers. For an overview, see Setting up licenses. Alternatively, you can generate licenses separately and merge the resulting license files.

Host requirements

Vocalizer runs on Linux and Windows hosts.

See the Release Notes for the supported versions.

Nuance currently tests and supports this product on hosts set to the en-US locale. Running on other locales might affect recognition accuracy and application performance.

User account requirements

You require these operating system accounts to install or remove Nuance speech products:

You must be root or a "sudoer" user. Processes can run under any user member of group "nuance."

Speech Suite services run under the user nuance and group nuance account. When you use startup scripts provided by Nuance, the processes will start as the nuance user (for example, the script Speech_Server/server/start or /etc/init.d/nuance-licmgr).

You must be logged in as Administrator. To run or stop Nuance products, you need an account with proper rights to directory NUANCE_DATA_DIR and subdirectories: modify, read, write. This account can be the user Administrator, a user in the group Administrators, or a standard user with these rights for the folder.

Network requirements

Nuance speech products support HTTP 1.1 and HTTPS. If your network uses secure HTTPS, you require these OpenSSL libraries:

  • On Windows, Vocalizer uses its own build of the libraries.
  • On Linux, Vocalizer uses the libraries supplied with the operating system. The OpenSSL libraries must be installed before installing Vocalizer. If OpenSSL is not present, the Vocalizer installation issues a warning.

    Vocalizer dynamically loads the OpenSSL libraries using the dlopen() function and the names libssl.so and libcrypto.so, which are typically symbolic links within /usr/lib to the most recent OpenSSL patch release on the system (typically /lib/libcrypto.so.version and /lib/libssl.so.version).

    See the Release Notes for the supported versions.

Voices

To run Vocalizer, you must install at least one voice on the Vocalizer host. Install the voice after installing Vocalizer. See Installing Vocalizer.