diff --git a/OS/01/start.sh b/OS/01/start.sh
index 8a4dcc6..0ff76f0 100755
--- a/OS/01/start.sh
+++ b/OS/01/start.sh
@@ -1,9 +1,11 @@
 ### SETTINGS
 
 # If ALL_LOCAL is False, we'll use OpenAI's services
-# If setting ALL_LOCAL to true, set the path to the WHISPER local model
+# else we use whisper.cpp and piper local models
 export ALL_LOCAL=False
-# export WHISPER_MODEL_PATH=...
+export WHISPER_MODEL_NAME="ggml-tiny.en.bin"
+
+# Set the OpenAI API key for OpenInterpreter to work
 # export OPENAI_API_KEY=sk-...
 
 # For TTS, we use the en_US-lessac-medium voice model by default
@@ -32,6 +34,11 @@ export LOG_LEVEL="INFO"
 ### 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)
diff --git a/OS/01/stt.py b/OS/01/stt.py
index aa14e24..48f0ed6 100644
--- a/OS/01/stt.py
+++ b/OS/01/stt.py
@@ -53,13 +53,14 @@ def run_command(command):
     return result.stdout, result.stderr
 
 def get_transcription_file(wav_file_path: str):
-    model_path = os.getenv("WHISPER_MODEL_PATH")
-    if not model_path:
-        raise EnvironmentError("WHISPER_MODEL_PATH environment variable is not set.")
+    whisper_rust_path = os.path.join(os.path.dirname(__file__), 'local_stt', 'whisper-rust')
+    model_name = os.getenv('WHISPER_MODEL_NAME')
+    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'),
-        '--model-path', model_path,
+        os.path.join(whisper_rust_path, 'whisper-rust'),
+        '--model-path', os.path.join(whisper_rust_path, model_name),
         '--file-path', wav_file_path
     ])