Merge branch 'main' into feature/dotenv

# Conflicts:
#	OS/01/start.sh

OS/01/device.py

@@ -189,8 +189,8 @@ async def websocket_communication(WS_URL):
   
 
         except Exception as e:
-            logging.exception(f"An error occurred during websocket communication. {e}")
+            logger.warn(f"An error occurred during websocket communication. {e}")
-            logging.info(f"Connecting to `{WS_URL}`...")
+            logger.info(f"Connecting to `{WS_URL}`...")
             await asyncio.sleep(2)
             
 

OS/01/start.sh

@@ -3,12 +3,16 @@ if [ ! -f ".env" ]; then
     echo "Error: .env file does not exist. To create one, see .env.example for an example."
     exit 1
 fi
-set -a; source .env; set +a
+set -a; source .env; set +az
-
 
 ### SETUP
 
 # if using local models, install the models / executables
+WHISPER_MODEL_URL="https://huggingface.co/ggerganov/whisper.cpp/resolve/main/"
+WHISPER_RUST_PATH="`pwd`/local_stt/whisper-rust"
+
+curl -OL "${WHISPER_MODEL_URL}${WHISPER_MODEL_NAME}" --output-dir ${WHISPER_RUST_PATH}
+
 if [[ "$ALL_LOCAL" == "True" ]]; then
     OS=$(uname -s)
     ARCH=$(uname -m)

OS/01/stt.py

@@ -56,13 +56,14 @@ def run_command(command):
     return result.stdout, result.stderr
 
 def get_transcription_file(wav_file_path: str):
-    model_path = os.getenv("WHISPER_MODEL_PATH")
+    whisper_rust_path = os.path.join(os.path.dirname(__file__), 'local_stt', 'whisper-rust')
-    if not model_path:
+    model_name = os.getenv('WHISPER_MODEL_NAME')
-        raise EnvironmentError("WHISPER_MODEL_PATH environment variable is not set.")
+    if not model_name:
+        raise EnvironmentError("WHISPER_MODEL_NAME environment variable is not set.")
 
     output, error = run_command([
-        os.path.join(os.path.dirname(__file__), 'local_stt', 'whisper-rust', 'whisper-rust'),
+        os.path.join(whisper_rust_path, 'whisper-rust'),
-        '--model-path', model_path,
+        '--model-path', os.path.join(whisper_rust_path, model_name),
         '--file-path', wav_file_path
     ])
 

README.md

@@ -27,8 +27,6 @@ python -m pip install -r requirements.txt
 ```
 NB: Depending on your local Python version, you may run into [this issue↗](https://github.com/TaylorSMarks/playsound/issues/150) installing playsound. Workarounds are provided in the issue.
 
-If you want to run local speech-to-text from whisper, download the GGML Whisper model from [Huggingface](https://huggingface.co/ggerganov/whisper.cpp). Then in `OS/01/start.sh`, set `ALL_LOCAL=TRUE` and set `WHISPER_MODEL_PATH` to the path of the model.
-
 ## Usage
 
 ```bash
@@ -36,6 +34,7 @@ cd OS/01
 bash start.sh
 ```
 
+If you want to run local text-to-speech and speech-to-text, set `ALL_LOCAL` in your `OS/01/.env` config to True. This will use the [whisper.cpp](https://github.com/ggerganov/whisper.cpp) and [Piper](https://github.com/rhasspy/piper) models.
 <br>
 
 ## Background