modular-voice-agent-sdk 1.0.0

package/scripts/setup-binaries.sh

@@ -0,0 +1,159 @@
+#!/bin/bash
+#
+# Setup native binaries for modular-voice-agent-sdk
+# Binaries are stored in ~/.cache/mvas/bin/
+#
+# Usage: npx mvas setup --binaries-only
+#
+
+set -e
+
+# Use global cache directory
+CACHE_DIR="${MVAS_CACHE:-$HOME/.cache/mvas}"
+BIN_DIR="$CACHE_DIR/bin"
+
+echo "Modular Voice Agent SDK - Binary Setup"
+echo "======================================="
+echo "Binary directory: $BIN_DIR"
+echo ""
+
+mkdir -p "$BIN_DIR"
+
+cd "$BIN_DIR"
+
+# Detect platform
+OS="$(uname -s)"
+ARCH="$(uname -m)"
+
+# --- whisper.cpp ---
+echo "==> Setting up whisper-cli..."
+if [ -L "whisper-cli" ] || [ -f "whisper-cli" ]; then
+  echo "    Already exists, skipping."
+else
+  # Try to find whisper-cli from Homebrew or PATH
+  if command -v whisper-cli &> /dev/null; then
+    WHISPER_PATH="$(command -v whisper-cli)"
+    ln -s "$WHISPER_PATH" whisper-cli
+    echo "    Linked to $WHISPER_PATH"
+  elif [ -f "/opt/homebrew/bin/whisper-cli" ]; then
+    ln -s /opt/homebrew/bin/whisper-cli whisper-cli
+    echo "    Linked to /opt/homebrew/bin/whisper-cli"
+  elif [ -f "/usr/local/bin/whisper-cli" ]; then
+    ln -s /usr/local/bin/whisper-cli whisper-cli
+    echo "    Linked to /usr/local/bin/whisper-cli"
+  else
+    echo "    ⚠️  whisper-cli not found. Install with: brew install whisper-cpp"
+  fi
+fi
+
+# --- llama.cpp (llama-completion) ---
+echo ""
+echo "==> Setting up llama-completion..."
+if [ -L "llama-completion" ] || [ -f "llama-completion" ]; then
+  echo "    Already exists, skipping."
+else
+  # Try to find llama-completion from PATH or common locations
+  if command -v llama-completion &> /dev/null; then
+    LLAMA_PATH="$(command -v llama-completion)"
+    ln -s "$LLAMA_PATH" llama-completion
+    echo "    Linked to $LLAMA_PATH"
+  elif [ -f "/opt/homebrew/bin/llama-completion" ]; then
+    ln -s /opt/homebrew/bin/llama-completion llama-completion
+    echo "    Linked to /opt/homebrew/bin/llama-completion"
+  elif [ -f "/usr/local/bin/llama-completion" ]; then
+    ln -s /usr/local/bin/llama-completion llama-completion
+    echo "    Linked to /usr/local/bin/llama-completion"
+  else
+    echo "    ⚠️  llama-completion not found."
+    echo "       Install with: brew install llama.cpp"
+    echo "       Or build from source: git clone https://github.com/ggerganov/llama.cpp && cd llama.cpp && make"
+  fi
+fi
+
+# --- sherpa-onnx (TTS) ---
+echo ""
+echo "==> Setting up sherpa-onnx for $OS/$ARCH..."
+
+if [ -f "sherpa-onnx-offline-tts" ]; then
+  echo "    Already exists, skipping."
+else
+  case "$OS" in
+    Darwin)
+      # macOS - universal binary (arm64 + x86_64)
+      SHERPA_URL="https://github.com/k2-fsa/sherpa-onnx/releases/download/v1.12.20/sherpa-onnx-v1.12.20-osx-universal2-shared.tar.bz2"
+      SHERPA_DIR="sherpa-onnx-v1.12.20-osx-universal2-shared"
+      ;;
+    Linux)
+      if [ "$ARCH" = "x86_64" ]; then
+        SHERPA_URL="https://github.com/k2-fsa/sherpa-onnx/releases/download/v1.12.20/sherpa-onnx-v1.12.20-linux-x64-shared.tar.bz2"
+        SHERPA_DIR="sherpa-onnx-v1.12.20-linux-x64-shared"
+      elif [ "$ARCH" = "aarch64" ]; then
+        SHERPA_URL="https://github.com/k2-fsa/sherpa-onnx/releases/download/v1.12.20/sherpa-onnx-v1.12.20-linux-aarch64-shared.tar.bz2"
+        SHERPA_DIR="sherpa-onnx-v1.12.20-linux-aarch64-shared"
+      else
+        echo "    Unsupported Linux architecture: $ARCH"
+        exit 1
+      fi
+      ;;
+    *)
+      echo "    Unsupported OS: $OS"
+      echo "    Download manually from: https://github.com/k2-fsa/sherpa-onnx/releases"
+      exit 1
+      ;;
+  esac
+
+  echo "    Downloading from $SHERPA_URL..."
+  curl -L --progress-bar -o sherpa-onnx.tar.bz2 "$SHERPA_URL"
+  tar -xjf sherpa-onnx.tar.bz2
+
+  # Only copy the TTS binary we need (not all the extras)
+  cp "$SHERPA_DIR/bin/sherpa-onnx-offline-tts" .
+  mkdir -p lib
+  mv "$SHERPA_DIR/lib/"* lib/
+
+  # Fix library paths on macOS so binary can find libs in ./lib/
+  if [ "$OS" = "Darwin" ]; then
+    echo "    Fixing library paths..."
+    for lib in lib/*.dylib; do
+      libname=$(basename "$lib")
+      install_name_tool -change "@rpath/$libname" "@executable_path/lib/$libname" sherpa-onnx-offline-tts 2>/dev/null || true
+    done
+  fi
+
+  # Cleanup
+  rm -rf "$SHERPA_DIR" sherpa-onnx.tar.bz2
+
+  echo "    Done!"
+fi
+
+# ============ Summary ============
+
+echo ""
+echo "============================================================"
+echo "Binary setup complete!"
+echo "============================================================"
+echo ""
+echo "Binaries location: $BIN_DIR"
+echo ""
+ls -lh "$BIN_DIR" 2>/dev/null || true
+
+# Check for missing dependencies
+MISSING_BREW=""
+[ ! -L "$BIN_DIR/whisper-cli" ] && [ ! -f "$BIN_DIR/whisper-cli" ] && MISSING_BREW="$MISSING_BREW whisper-cpp"
+[ ! -L "$BIN_DIR/llama-completion" ] && [ ! -f "$BIN_DIR/llama-completion" ] && MISSING_BREW="$MISSING_BREW llama.cpp"
+
+if [ -n "$MISSING_BREW" ]; then
+  echo ""
+  echo "⚠️  Missing dependencies:"
+  echo "  brew install$MISSING_BREW"
+  echo ""
+  echo "Then re-run: npx mvas setup --binaries-only"
+else
+  echo ""
+  echo "✅ All binaries set up!"
+fi
+
+echo ""
+echo "You can customize the cache location with:"
+echo "  export MVAS_CACHE=/path/to/cache"
+

package/scripts/setup.sh

@@ -0,0 +1,201 @@
+#!/bin/bash
+#
+# Setup script - downloads models and binaries for native backends
+# Files are stored in ~/.cache/mvas/
+#
+# Usage: npx mvas setup
+#
+
+set -e
+
+# Use global cache directory
+CACHE_DIR="${MVAS_CACHE:-$HOME/.cache/mvas}"
+MODELS_DIR="$CACHE_DIR/models"
+BIN_DIR="$CACHE_DIR/bin"
+
+echo "Modular Voice Agent SDK Setup"
+echo "=============================="
+echo "Cache directory: $CACHE_DIR"
+echo ""
+
+mkdir -p "$MODELS_DIR"
+mkdir -p "$BIN_DIR"
+
+# ============ Models ============
+
+cd "$MODELS_DIR"
+
+# Whisper model (whisper.cpp format - Large V3 Turbo quantized)
+echo "==> Downloading Whisper model (~850MB)..."
+if [ -f "whisper-large-v3-turbo-q8.bin" ]; then
+  echo "    Already exists, skipping."
+else
+  curl -L --progress-bar -o whisper-large-v3-turbo-q8.bin \
+    "https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo-q8_0.bin"
+fi
+
+# LLM model (GGUF format)
+echo ""
+echo "==> Downloading SmolLM2 model (~1GB)..."
+if [ -f "smollm2-1.7b-instruct-q4_k_m.gguf" ]; then
+  echo "    Already exists, skipping."
+else
+  curl -L --progress-bar -o smollm2-1.7b-instruct-q4_k_m.gguf \
+    "https://huggingface.co/HuggingFaceTB/SmolLM2-1.7B-Instruct-GGUF/resolve/main/smollm2-1.7b-instruct-q4_k_m.gguf"
+fi
+
+# TTS model (sherpa-onnx compatible Piper model)
+echo ""
+echo "==> Downloading TTS model (~60MB)..."
+if [ -d "vits-piper-en_US-lessac-medium" ]; then
+  echo "    Already exists, skipping."
+else
+  curl -L --progress-bar -o vits-piper-en_US-lessac-medium.tar.bz2 \
+    "https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-lessac-medium.tar.bz2"
+  tar -xjf vits-piper-en_US-lessac-medium.tar.bz2
+  rm vits-piper-en_US-lessac-medium.tar.bz2
+fi
+
+# ============ Binaries ============
+
+cd "$BIN_DIR"
+
+# Detect platform
+OS="$(uname -s)"
+ARCH="$(uname -m)"
+
+# --- whisper.cpp ---
+echo ""
+echo "==> Setting up whisper-cli..."
+if [ -L "whisper-cli" ] || [ -f "whisper-cli" ]; then
+  echo "    Already exists, skipping."
+else
+  # Try to find whisper-cli from Homebrew or PATH
+  if command -v whisper-cli &> /dev/null; then
+    WHISPER_PATH="$(command -v whisper-cli)"
+    ln -s "$WHISPER_PATH" whisper-cli
+    echo "    Linked to $WHISPER_PATH"
+  elif [ -f "/opt/homebrew/bin/whisper-cli" ]; then
+    ln -s /opt/homebrew/bin/whisper-cli whisper-cli
+    echo "    Linked to /opt/homebrew/bin/whisper-cli"
+  elif [ -f "/usr/local/bin/whisper-cli" ]; then
+    ln -s /usr/local/bin/whisper-cli whisper-cli
+    echo "    Linked to /usr/local/bin/whisper-cli"
+  else
+    echo "    ⚠️  whisper-cli not found. Install with: brew install whisper-cpp"
+  fi
+fi
+
+# --- llama.cpp (llama-completion) ---
+echo ""
+echo "==> Setting up llama-completion..."
+if [ -L "llama-completion" ] || [ -f "llama-completion" ]; then
+  echo "    Already exists, skipping."
+else
+  # Try to find llama-completion from PATH or common locations
+  if command -v llama-completion &> /dev/null; then
+    LLAMA_PATH="$(command -v llama-completion)"
+    ln -s "$LLAMA_PATH" llama-completion
+    echo "    Linked to $LLAMA_PATH"
+  elif [ -f "/opt/homebrew/bin/llama-completion" ]; then
+    ln -s /opt/homebrew/bin/llama-completion llama-completion
+    echo "    Linked to /opt/homebrew/bin/llama-completion"
+  elif [ -f "/usr/local/bin/llama-completion" ]; then
+    ln -s /usr/local/bin/llama-completion llama-completion
+    echo "    Linked to /usr/local/bin/llama-completion"
+  else
+    echo "    ⚠️  llama-completion not found."
+    echo "       Install with: brew install llama.cpp"
+    echo "       Or build from source: git clone https://github.com/ggerganov/llama.cpp && cd llama.cpp && make"
+  fi
+fi
+
+# --- sherpa-onnx (TTS) ---
+echo ""
+echo "==> Setting up sherpa-onnx for $OS/$ARCH..."
+
+if [ -f "sherpa-onnx-offline-tts" ]; then
+  echo "    Already exists, skipping."
+else
+  case "$OS" in
+    Darwin)
+      # macOS - universal binary (arm64 + x86_64)
+      SHERPA_URL="https://github.com/k2-fsa/sherpa-onnx/releases/download/v1.12.20/sherpa-onnx-v1.12.20-osx-universal2-shared.tar.bz2"
+      SHERPA_DIR="sherpa-onnx-v1.12.20-osx-universal2-shared"
+      ;;
+    Linux)
+      if [ "$ARCH" = "x86_64" ]; then
+        SHERPA_URL="https://github.com/k2-fsa/sherpa-onnx/releases/download/v1.12.20/sherpa-onnx-v1.12.20-linux-x64-shared.tar.bz2"
+        SHERPA_DIR="sherpa-onnx-v1.12.20-linux-x64-shared"
+      elif [ "$ARCH" = "aarch64" ]; then
+        SHERPA_URL="https://github.com/k2-fsa/sherpa-onnx/releases/download/v1.12.20/sherpa-onnx-v1.12.20-linux-aarch64-shared.tar.bz2"
+        SHERPA_DIR="sherpa-onnx-v1.12.20-linux-aarch64-shared"
+      else
+        echo "    Unsupported Linux architecture: $ARCH"
+        exit 1
+      fi
+      ;;
+    *)
+      echo "    Unsupported OS: $OS"
+      echo "    Download manually from: https://github.com/k2-fsa/sherpa-onnx/releases"
+      exit 1
+      ;;
+  esac
+
+  echo "    Downloading from $SHERPA_URL..."
+  curl -L --progress-bar -o sherpa-onnx.tar.bz2 "$SHERPA_URL"
+  tar -xjf sherpa-onnx.tar.bz2
+
+  # Only copy the TTS binary we need (not all the extras)
+  cp "$SHERPA_DIR/bin/sherpa-onnx-offline-tts" .
+  mkdir -p lib
+  mv "$SHERPA_DIR/lib/"* lib/
+
+  # Fix library paths on macOS so binary can find libs in ./lib/
+  if [ "$OS" = "Darwin" ]; then
+    echo "    Fixing library paths..."
+    for lib in lib/*.dylib; do
+      libname=$(basename "$lib")
+      install_name_tool -change "@rpath/$libname" "@executable_path/lib/$libname" sherpa-onnx-offline-tts 2>/dev/null || true
+    done
+  fi
+
+  # Cleanup
+  rm -rf "$SHERPA_DIR" sherpa-onnx.tar.bz2
+
+  echo "    Done!"
+fi
+
+# ============ Summary ============
+
+echo ""
+echo "============================================================"
+echo "Setup complete!"
+echo "============================================================"
+echo ""
+echo "Cache location: $CACHE_DIR"
+echo ""
+echo "Models:"
+ls -lh "$MODELS_DIR"
+echo ""
+echo "Binaries:"
+ls -lh "$BIN_DIR"
+
+# Check for missing dependencies
+MISSING_BREW=""
+[ ! -L "$BIN_DIR/whisper-cli" ] && [ ! -f "$BIN_DIR/whisper-cli" ] && MISSING_BREW="$MISSING_BREW whisper-cpp"
+[ ! -L "$BIN_DIR/llama-completion" ] && [ ! -f "$BIN_DIR/llama-completion" ] && MISSING_BREW="$MISSING_BREW llama.cpp"
+
+if [ -n "$MISSING_BREW" ]; then
+  echo ""
+  echo "⚠️  Missing dependencies:"
+  echo "  brew install$MISSING_BREW"
+  echo ""
+  echo "Then re-run: npx mvas setup"
+else
+  echo ""
+  echo "✅ All dependencies installed!"
+  echo ""
+  echo "You can customize the cache location with:"
+  echo "  export MVAS_CACHE=/path/to/cache"
+fi