updated stale links, added simple run script

.gitignore

@@ -1 +1,6 @@
 __pycache__
+separated_results/
+checkpoints/
+.idea/
+_zz_*
+/.DS_Store

run.sh

@@ -0,0 +1,83 @@
+#!/bin/bash
+
+mkdir -p "./checkpoints"
+
+MODEL1_CKPT="./checkpoints/ss_model=resunet30,querynet=at_soft,data=full,devices=8,step=1000000.ckpt"
+MODEL1_YAML="./checkpoints/ss_model=resunet30,querynet=at_soft,data=full.yaml"
+MODEL2_CKPT="./checkpoints/ss_model=resunet30,querynet=emb,data=balanced,devices=1,steps=1000000.ckpt"
+
+URL_MODEL1_CKPT="https://huggingface.co/RSNuts/Universal_Source_Separation/resolve/main/uss_material/ss_model%3Dresunet30%2Cquerynet%3Dat_soft%2Cdata%3Dfull%2Cdevices%3D8%2Cstep%3D1000000.ckpt"
+URL_MODEL1_YAML="https://huggingface.co/RSNuts/Universal_Source_Separation/resolve/main/uss_material/ss_model%3Dresunet30%2Cquerynet%3Dat_soft%2Cdata%3Dfull.yaml?download=true"
+URL_MODEL2_CKPT="https://huggingface.co/RSNuts/Universal_Source_Separation/resolve/main/uss_material/ss_model%3Dresunet30%2Cquerynet%3Demb%2Cdata%3Dbalanced%2Cdevices%3D1%2Csteps%3D1000000.ckpt"
+
+# ---------------------------
+#  AUDIO INPUT HANDLING
+# ---------------------------
+
+DEFAULT_AUDIO="./resources/harry_potter.flac"
+INPUT_FILE="$1"
+
+if [ -z "$INPUT_FILE" ]; then
+    echo "No input file provided. Using default: $DEFAULT_AUDIO"
+    AUDIO_FILE="$DEFAULT_AUDIO"
+else
+    BASENAME=$(basename "$INPUT_FILE")
+    NAME="${BASENAME%.*}"
+    EXT="${BASENAME##*.}"
+    EXT_LOWER=$(echo "$EXT" | tr '[:upper:]' '[:lower:]')
+
+    OUTFILE="/tmp/${NAME}.flac"
+    mkdir -p /tmp
+
+    # Audio-only formats we accept directly:
+    case "$EXT_LOWER" in
+        flac)
+            echo "Copying FLAC file to /tmp..."
+            cp "$INPUT_FILE" "$OUTFILE"
+            AUDIO_FILE="$OUTFILE"
+            ;;
+        mp3|wav|ogg|m4a|aac)
+            echo "Converting audio file $INPUT_FILE → $OUTFILE"
+            ffmpeg -y -i "$INPUT_FILE" -ac 2 -sample_fmt s16 -ar 44100 "$OUTFILE"
+            AUDIO_FILE="$OUTFILE"
+            ;;
+        *)
+            # Assume any unknown extension *might* be video — FFmpeg will confirm
+            echo "Attempting to extract audio from video file $INPUT_FILE → $OUTFILE"
+            ffmpeg -y -i "$INPUT_FILE" -vn -ac 2 -sample_fmt s16 -ar 44100 "$OUTFILE"
+
+            if [ $? -ne 0 ]; then
+                echo "Error: FFmpeg cannot process this file type: $INPUT_FILE"
+                exit 1
+            fi
+
+            AUDIO_FILE="$OUTFILE"
+            ;;
+    esac
+fi
+
+# ---------------------------
+#  CHECKPOINT DOWNLOAD
+# ---------------------------
+
+if [ ! -f "$MODEL1_CKPT" ]; then
+    wget -O "$MODEL1_CKPT" "$URL_MODEL1_CKPT"
+fi
+
+if [ ! -f "$MODEL1_YAML" ]; then
+    wget -O "$MODEL1_YAML" "$URL_MODEL1_YAML"
+fi
+
+if [ ! -f "$MODEL2_CKPT" ]; then
+    wget -O "$MODEL2_CKPT" "$URL_MODEL2_CKPT"
+fi
+
+# ---------------------------
+#  RUN INFERENCE
+# ---------------------------
+
+PYTHONPATH=~/source/uss CUDA_VISIBLE_DEVICES=0 python uss/inference.py \
+    --audio_path="$AUDIO_FILE" \
+    --levels 1 2 3 \
+    --config_yaml="$MODEL1_YAML" \
+    --checkpoint_path="$MODEL1_CKPT"

uss/config.py

@@ -7,12 +7,12 @@
 csv_paths_dict = {
     "class_labels_indices.csv": {
         "path": Path(Path.home(), ".cache/uss/metadata/class_labels_indices.csv"),
-        "remote_path": "https://sandbox.zenodo.org/record/1186898/files/class_labels_indices.csv?download=1",
+        "remote_path": "https://huggingface.co/RSNuts/Universal_Source_Separation/resolve/main/uss_material/class_labels_indices.csv?download=true",
         "size": 14675,
     },
     "ontology.csv": {
         "path": Path(Path.home(), ".cache/uss/metadata/ontology.json"),
-        "remote_path": "https://sandbox.zenodo.org/record/1186898/files/ontology.json?download=1",
+        "remote_path": "https://huggingface.co/RSNuts/Universal_Source_Separation/resolve/main/uss_material/ontology.json?download=true",
         "size": 342780,
     },
 }