summaryrefslogblamecommitdiff
path: root/makima/sh/download-models.sh
blob: 7aecefe0491fea7d8fcbe35e01a4ebc8377b41fe (plain) (tree)
1
2
3
4
5
6
7
8
9
10
11





                                       
                   



                                   
                     








                                                               
                                        
                    
                                                            


                                        













                                                               
                                                                    
        
                                               




                                           









                                                                                

                          
















                                                                                                     





                                             
#!/bin/bash
set -e

MODELS_DIR="${MODELS_DIR:-/app/models}"
MODEL_BASE_URL="${MODEL_BASE_URL:-}"

# Model directories
PARAKEET_DIR="parakeet-tdt-0.6b-v3"
EOU_DIR="realtime_eou_120m-v1-onnx"
DIARIZATION_DIR="diarization"

download_from_url() {
    local name=$1
    local url=$2
    local dest="$MODELS_DIR/$name"

    if [ -d "$dest" ] && [ "$(ls -A $dest 2>/dev/null)" ]; then
        echo "Model $name already exists, skipping..."
        return 0
    fi

    echo "Downloading $name from URL..."
    mkdir -p "$dest"
    curl -L "$url" | tar -xz -C "$dest" --strip-components=1
    echo "Downloaded $name successfully"
}

download_from_hf() {
    local dest=$1
    local repo=$2
    local include=${3:-}

    if [ -d "$dest" ] && [ "$(ls -A $dest 2>/dev/null)" ]; then
        echo "Model $dest already exists, skipping..."
        return 0
    fi

    echo "Downloading from Hugging Face ($repo)..."
    mkdir -p "$dest"

    if [ -n "$include" ]; then
        hf download "$repo" --include "$include" --local-dir "$dest"
    else
        hf download "$repo" --local-dir "$dest"
    fi

    echo "Downloaded to $dest successfully"
}

# Check if models exist
check_models_exist() {
    [ -d "$MODELS_DIR/$PARAKEET_DIR" ] && \
    [ -d "$MODELS_DIR/$EOU_DIR" ] && \
    [ -f "$MODELS_DIR/$DIARIZATION_DIR/diar_streaming_sortformer_4spk-v2.onnx" ]
}

if check_models_exist; then
    echo "All models present"
else
    mkdir -p "$MODELS_DIR"

    if [ -n "$MODEL_BASE_URL" ]; then
        echo "Downloading models from custom URL..."
        download_from_url "$PARAKEET_DIR" "${MODEL_BASE_URL}/parakeet-tdt-0.6b-v3.tar.gz"
        download_from_url "$EOU_DIR" "${MODEL_BASE_URL}/realtime_eou_120m-v1-onnx.tar.gz"
        download_from_url "$DIARIZATION_DIR" "${MODEL_BASE_URL}/diarization.tar.gz"
    else
        echo "Downloading models from Hugging Face..."

        # Parakeet TDT from istupakov/parakeet-tdt-0.6b-v3-onnx
        download_from_hf "$MODELS_DIR/$PARAKEET_DIR" "istupakov/parakeet-tdt-0.6b-v3-onnx"

        # EOU model from altunenes/parakeet-rs (subdirectory)
        download_from_hf "$MODELS_DIR/$EOU_DIR" "altunenes/parakeet-rs" "realtime_eou_120m-v1-onnx/*"

        # Diarization model from altunenes/parakeet-rs (subdirectory)
        download_from_hf "$MODELS_DIR/$DIARIZATION_DIR" "altunenes/parakeet-rs" "diarization/*"
    fi

    echo "All models downloaded successfully"
fi

# Execute the main command
exec "$@"