Fetch every instrument when the image is built

Each instrument was downloaded from the Magenta bucket on Google Cloud Storage the first time a request picked it. The image now holds all five.

Dockerfile

@@ -48,5 +48,12 @@ RUN /usr/local/bin/python3.10 -m venv "$FRONTEND_VENV" \
 COPY backend_worker.py frontend_app.py start.sh ./
 RUN chmod +x /app/start.sh
 
+# Every instrument's checkpoint, from the Magenta bucket the backend would otherwise reach on the
+# first request for each instrument. Set here, after the environments, so that their layers stay
+# cached.
+ENV DDSP_MODELS_DIR=/opt/ddsp_pretrained
+RUN "$BACKEND_PYTHON" -c "import backend_worker as w; [w._ensure_model(m) for m in w.PRETRAINED_MODELS]" \
+    && chmod -R a+rX "$DDSP_MODELS_DIR"
+
 EXPOSE 7860
 CMD ["/app/start.sh"]

backend_worker.py

@@ -39,7 +39,9 @@ SAMPLE_RATE = 16000
 PRETRAINED_MODELS = ["Violin", "Flute", "Flute2", "Trumpet", "Tenor_Saxophone"]
 _GCS_BUCKET = "ddsp"
 _GCS_PREFIX = "models/timbre_transfer_colab/2021-07-08"
-_MODELS_DIR = "/tmp/ddsp_pretrained"
+# The image fetches every instrument when it is built (see the Dockerfile), so no request waits on
+# Google Cloud Storage. A directory without one still downloads it on first use.
+_MODELS_DIR = os.environ.get("DDSP_MODELS_DIR", "/tmp/ddsp_pretrained")
 os.makedirs(_MODELS_DIR, exist_ok=True)