This section is meant to get you started using Cubic Server via a Docker image.
It is not necessary to go through through these steps to call the demo server for evaluation purposes, only to run Cubic Server on premises. The demo server can be found at https://demo.cobaltspeech.com/cubic.
The SDK communicates with a Cubic Server instance using gRPC. Cobalt distributes a docker image that contains the cubicsvr binary and model files.
Contact Cobalt to get a link to the image file in AWS S3. This link will expire in two weeks, so be sure to download the file to your own server.
Download with the AWS CLI if you have it, or with curl:
URL="the url sent by Cobalt"
IMAGE_NAME="name you want to give the file (should end with the same extension as the url, usually bz2)"
curl $URL -L -o $IMAGE_NAME
Load the docker image
docker load < $IMAGE_NAME
This will output the name of the image (e.g. cubicsvr-demo-en_us-16).
Start the cubic service
docker run -p 2727:2727 -p 8080:8080 --name cobalt -v /path/on/host/usage-log:/usage cubicsvr-demo-en_us-16
That will start listening for grpc commands on port 2727 and http requests on 8080, and will stream the debug log to stdout. (You can replace --name cobalt with whatever name you want. That just provides a way to refer back to the currently running container.)
The command also specifies the bind-mount to which usage data is being written so you can track the amount of audio Cubic Server has processed, and submit the usage log to Cobalt if you have a volume-based license. If you have a volume-based license, you MUST provide this bind-mount so the usage data will persist even when the container is restarted; Cubic Server will not start if that location cannot be written to.
Verify the service is running by calling
curl http://localhost:8080/api/version
If you want to explore the package to see the model files etc, call the following to open the bash terminal on the previously run image. Model files are located in the /model directory.
docker exec -it cobalt bash