Less Memory Intensive Models

NOTE: As of LumenVox version 9.1, the following information is obsolete. We are leaving it as a reference for users of older versions of the software. For similar functionality in releases since 9.1, please see Different Acoustic Model Resolutions

If you have memory limitations on the computer used to run the Speech Engine, you may find it helpful to use our smaller language models. Though less robust -- they are less accurate in some domains -- than our default models, the less memory intensive models are significantly smaller and require less memory to use.

These models are located by in the Engine installation directory, then in Lang\Dict\less_intensive_memory_models. On Windows, the default location would be C:\Program Files\LumenVox\Engine\Lang\Dict\less_intensive_memory_models\ and on Linux the default location would be /opt/lumenvox/Engine_7.5/Lang/Dict/less_intensive_memory_models/.

The default models are located in the Dict directory. To use the less intensive models:

1. Stop the Speech Engine service.
2. Back up the default models by moving them to a different location.
3. Copy the less intensive models to the Dict directory.
4. Restart the Engine.

If you ever desire to return to the larger models, simply stop the Engine and move the larger models back into the Dict directory.