diff options
| author | jwansek <eddie.atten.ea29@gmail.com> | 2024-05-10 12:44:12 +0100 |
|---|---|---|
| committer | jwansek <eddie.atten.ea29@gmail.com> | 2024-05-10 12:44:12 +0100 |
| commit | a67797e2b7565e96ba21a56558aa7fa5112c4a9f (patch) | |
| tree | c7a1af9852ca0465029da3fda4546f4bfbd3ab5d | |
| parent | e54316a3330b4531f070789c285c624e279a79f3 (diff) | |
| download | noetic-llama-a67797e2b7565e96ba21a56558aa7fa5112c4a9f.tar.gz noetic-llama-a67797e2b7565e96ba21a56558aa7fa5112c4a9f.zip | |
Updated README
| -rw-r--r-- | README.md | 8 |
1 files changed, 8 insertions, 0 deletions
@@ -3,6 +3,14 @@ `docker-compose` docker containers required for LCASTOR intent recognition and transcription with whisper. Currently, `whisperwrapper` needs to be run outside of docker due to problems passing through the microphone, which is why it is symlinked here. +If being run in a VM, set the whisper API url ROS parameter: + +`rosparam set /stt/whisper_api_url 192.168.122.1:9000` + +If debugging, you might need to manually make the node transcribe: + +`rostopic pub /stt/listening ollamamessages/WhisperListening "listening: True" -1` + Given a python library `capabilities` containing only functions, it inspects this library to generate an initial prompt to tell the LLM the possible functions. Then, with a ROS service call, these functions are called: |