Add model auto-download and English logging

README.md

@@ -75,9 +75,39 @@ The repository now includes a container setup for running the API in Docker.
 - `./models` -> `/models`
 
 Place your Whisper model inside `./models/whisper` or update `model_path` in `config.docker.json`.
+That path must contain the actual Hugging Face model files, not just an empty folder.
+
+If you prefer automatic bootstrap, set:
+
+```json
+"auto_download_missing_model": true,
+"model_repo_id": "openai/whisper-large-v3"
+```
+
+With the default Docker config, the container will automatically download that
+model into `/models/whisper` on first start if the directory is missing or incomplete.
+
+Expected example layout:
+
+```text
+models/
+  whisper/
+    config.json
+    generation_config.json
+    preprocessor_config.json
+    tokenizer.json
+    tokenizer_config.json
+    model.safetensors
+```
 
 ### Build and run with Docker Compose
 
+If your selected model requires authentication on Hugging Face, export a token first:
+
+```bash
+export HF_TOKEN=your_huggingface_token
+```
+
 ```bash
 docker compose up --build
 ```
@@ -93,6 +123,7 @@ http://localhost:5042
 ```bash
 docker build -t whisper-api-server .
 docker run --rm -p 5042:5042 \
+  -e HF_TOKEN="$HF_TOKEN" \
   -v "$(pwd)/config.docker.json:/app/config.docker.json:ro" \
   -v "$(pwd)/logs:/app/logs" \
   -v "$(pwd)/history:/app/history" \
@@ -106,6 +137,7 @@ The container image installs the same Python dependencies as the local setup, in
 
 ```bash
 docker run --rm --gpus all -p 5042:5042 \
+  -e HF_TOKEN="$HF_TOKEN" \
   -v "$(pwd)/config.docker.json:/app/config.docker.json:ro" \
   -v "$(pwd)/logs:/app/logs" \
   -v "$(pwd)/history:/app/history" \
@@ -121,6 +153,8 @@ The service is configured through the `config.json` file:
 {
     "service_port": 5042,
     "model_path": "/path/to/whisper/model",
+    "auto_download_missing_model": false,
+    "model_repo_id": "openai/whisper-large-v3",
     "language": "russian",
     "enable_history": true,
     "max_history_days": 30,
@@ -152,6 +186,8 @@ The service is configured through the `config.json` file:
 |-----------|-------------|
 | `service_port` | Port on which the service will run |
 | `model_path` | Path to the Whisper model directory |
+| `auto_download_missing_model` | Download the configured fallback model into `model_path` when the local directory is missing or incomplete |
+| `model_repo_id` | Hugging Face model repo to download when auto-download is enabled |
 | `language` | Language for transcription (e.g., "russian", "english") |
 | `enable_history` | Whether to save transcription history (true/false) |
 | `max_history_days` | Number of days to keep transcription history before rotation |