Installing the acoustic model

To install the acoustic model selected for your application, execute the installer am-VERSION.run. See the following example, using the Brazilian Portuguese model:

$ sudo ./am-ptbr-8k-1.0.6-rc.run
Verifying archive integrity...  100%   All good.
Uncompressing am-ptbr-8k-1.0.6  100%

Installing acoustic model (AM) [am-ptbr-8k-1.0.6]
WARNING: Setting as default language package
Success!

The installed acoustic model becomes the default language (message: ``WARNING: Setting as default language package), in other words, all speech recognition will use the installed acoustic model.

Warning

The CPQD ASR currently operates with only one acoustic model. It is impossible to run speech recognition in Portuguese and Spanish on the same ASR. Two ASR instances must be used, one for each language.

Should you wish to change the acoustic model, e.g. change the language or the sample rate, simply install the acoustic model you wish to use, restart the CPQD ASR, and it will become the default model for recognition.

Warning

When installing an acoustic model, make sure it is included in your license.

If you wish to install an acoustic model without making it the standard package, run the installer with the option --default false, as in the following example:

$ sudo ./am-ptbr-8k-1.0.6-rc.run -- --default false
Verifying archive integrity...  100%   All good.
Uncompressing am-ptbr-8k-1.0.6  100%

Installing acoustic model (AM) [am-ptbr-8k-1.0.6]
Success!