80ms
with randomly generated sentences between 40 and 50 characters.
g5g.xlarge
or larger EC2 machine for on-prem rime-tts GPU
t2.micro
or larger EC2 machine with 10 GB storage for on-prem API instance.Ubuntu Server 22.04
. If there is a limitation on your side not to use this distribution, please let us know. For reference, below are other Linux distributions supported by NVIDIA:
docker-compose.yml
file with your editor of choice to define the services and their configurations.
When running on Kubernetes, ensure thatStart docker compose:MODEL_URL
points tohttp://0.0.0.0:8080/invocations
instead of the Docker Compose service name.
8000
for http traffic, and port 8001
for websockets traffic.
You’ll need to permit outbound network traffic to http://optimize.rime.ai/usage
and http://optimize.rime.ai/license
to verify with our servers that you have an active on-prem licensing agreement and to register usage. Additionally, you’ll need access to quay.io, a container image repository platform, so you’ll need to allow outbound traffic to their servers on port 443
.
Note: Once the containers are started, expect 5 minutes delay for warm up before sending first tts requests.
8001
. For example ws://localhost:8001
which will be eqivalent to our cloud websockets-json api .