agentvibes 3.3.0 → 3.4.0

package/.claude/config/audio-effects.cfg

@@ -49,4 +49,4 @@ BMad Master|reverb 50 60 100 pitch -100|agentvibes_soft_flamenco_loop.mp3|0.30
 _party_mode|compand 0.3,1 6:-70,-60,-20|agent_vibes_dark_chill_step_loop.mp3|0.40
 |||
 # Default (no agent specified) - clean with Bachata background|||
-default|reverb 40 50 70|agentvibes_soft_flamenco_loop.mp3|0.30
+default|reverb 20 50 50|agentvibes_soft_flamenco_loop.mp3|0.30

package/.claude/config/background-music-position.txt

@@ -1,26 +1 @@
-agent-vibes-dark-chill-step.mp3:49.006169
-Agent Vibes Japanese City Pop v1.mp3:29.392744
-Agent Vibes ChillWave v2.mp3:22.154467
-Agent Vibes Bossa Nova v2.mp3:23.733424
-Agent Vibes Tabla Dream Pop v1.mp3:19.101043
-Agent Vibes Hawaiian slack key guitar v2.mp3:36.381950
-AgentVibes Soft Flamenco.mp3:23.160000
-Agent Vibes Arabic v2.mp3:21.922268
-Agent Vibes Goa Trance v2.mp3:55.953741
-Agent Vibes Ganawa Ambient v2.mp3:39.680205
-Agent Vibes Celtic Harp v1.mp3:42.190476
-Agent Vibes Harpsichord v2.mp3:21.739410
-Agent Vibes Japanese City Pop v1-loop.mp3:13.917551
-Agent Vibes Hawaiian slack key guitar v2-loop.mp3:12.977143
-Agent Vibes Ganawa Ambient v2-loop.mp3:.00000000000000000002815996
-Agent Vibes Tabla Dream Pop v1-loop.mp3:.00000000000000000009067943
-Agent Vibes ChillWave v2-loop.mp3:.00000000000000000007080511
-Agent Vibes Harpsichord v2-loop.mp3:.00000000000000000013140818
-agent_vibes_japanese_city_pop_v1_loop.mp3:6.054512
-agent_vibes_bossa_nova_v2_loop.mp3:5.369524
-agent_vibes_salsa_v2_loop.mp3:9.972790
-agent_vibes_cumbia_v1_loop.mp3:5.717823
-agent_vibes_arabic_v2_loop.mp3:.00000000000000000006132724
-agent_vibes_chillwave_v2_loop.mp3:14.628390
-agent_vibes_bachata_v1_loop.mp3:.00000000000000000005344000
-agentvibes_soft_flamenco_loop.mp3:.00000000000000000006934441
+agentvibes_soft_flamenco_loop.mp3:9.624036

package/.claude/github-star-reminder.txt

@@ -1 +1 @@
-20260205
+20260210

package/.claude/hooks/play-tts-soprano.sh

@@ -0,0 +1,320 @@
+#!/usr/bin/env bash
+#
+# File: .claude/hooks/play-tts-soprano.sh
+#
+# AgentVibes - Finally, your AI Agents can Talk Back! Text-to-Speech WITH personality for AI Assistants!
+# Website: https://agentvibes.org
+# Repository: https://github.com/paulpreibisch/AgentVibes
+#
+# Co-created by Paul Preibisch with Claude AI
+# Copyright (c) 2025 Paul Preibisch
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+# DISCLAIMER: This software is provided "AS IS", WITHOUT WARRANTY OF ANY KIND,
+# express or implied. Use at your own risk. See the Apache License for details.
+#
+# ---
+#
+# @fileoverview Soprano TTS Provider Implementation - Free, local, neural-quality TTS
+# @context Provides ultra-lightweight on-device neural TTS via Soprano (80M params)
+# @architecture Implements provider interface contract with 3 synthesis modes (WebUI/API/CLI)
+# @dependencies soprano-tts (pip), soprano-gradio-synth.py, ffmpeg (optional padding), audio players
+# @entrypoints Called by play-tts.sh router when provider=soprano
+# @patterns Provider contract: text/voice → audio file path, auto-mode detection, Gradio SSE protocol
+# @related play-tts.sh, soprano-gradio-synth.py, provider-manager.sh, GitHub Issue #94
+#
+# Supports three modes (auto-detected in priority order):
+#   1. WebUI mode: Gradio WebUI running (soprano-webui), uses Python helper
+#   2. API mode: OpenAI-compatible server (uvicorn soprano.server:app), uses curl
+#   3. CLI mode: Direct `soprano` command — reloads model each call (slowest)
+#
+# Environment variables:
+#   SOPRANO_PORT    — WebUI/API port (default: 7860)
+#   SOPRANO_DEVICE  — Device for CLI mode: auto|cuda|cpu|mps (default: auto)
+#
+
+# Fix locale warnings
+export LC_ALL=C
+
+TEXT="$1"
+VOICE_OVERRIDE="$2"  # Ignored — Soprano has a single voice, kept for provider contract
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+source "$SCRIPT_DIR/audio-cache-utils.sh"
+
+SOPRANO_PORT="${SOPRANO_PORT:-7860}"
+SOPRANO_DEVICE="${SOPRANO_DEVICE:-auto}"
+
+# @function validate_inputs
+# @intent Check required parameters
+# @why Fail fast with clear errors if inputs missing
+# @exitcode 1=missing text
+if [[ -z "$TEXT" ]]; then
+  echo "Usage: $0 \"text to speak\" [voice_override]"
+  exit 1
+fi
+
+# @function check_webui_server
+# @intent Detect if Soprano Gradio WebUI is reachable
+# @why WebUI mode keeps model in memory for fastest repeated synthesis
+# @returns exitcode 0=reachable, 1=not reachable
+check_webui_server() {
+  curl -sf --max-time 2 "http://127.0.0.1:${SOPRANO_PORT}/gradio_api/info" -o /dev/null 2>/dev/null ||
+  curl -sf --max-time 2 "http://127.0.0.1:${SOPRANO_PORT}/info" -o /dev/null 2>/dev/null
+}
+
+# @function check_api_server
+# @intent Detect if Soprano OpenAI-compatible API server is reachable
+# @why API mode is simpler than WebUI (direct WAV response, no SSE polling)
+# @returns exitcode 0=reachable, 1=not reachable
+check_api_server() {
+  curl -sf --max-time 2 "http://127.0.0.1:${SOPRANO_PORT}/v1/audio/speech" \
+    -H "Content-Type: application/json" \
+    -d '{"input":"test"}' -o /dev/null 2>/dev/null
+}
+
+# @function check_soprano_available
+# @intent Verify at least one synthesis mode is available
+# @why Provide helpful installation instructions if nothing works
+# @exitcode 2=soprano not installed and no server running
+if ! command -v soprano &>/dev/null && ! check_webui_server && ! check_api_server; then
+  echo "❌ Error: Soprano TTS not installed and no server running on port $SOPRANO_PORT"
+  echo ""
+  echo "Install:  pip install soprano-tts"
+  echo "  (GPU):  pip install soprano-tts[lmdeploy]"
+  echo ""
+  echo "Start WebUI:  soprano-webui"
+  echo "Start API:    uvicorn soprano.server:app --host 127.0.0.1 --port $SOPRANO_PORT"
+  exit 2
+fi
+
+# @function determine_audio_directory
+# @intent Find appropriate directory for audio file storage
+# @why Supports project-local and global storage
+# @returns Sets $AUDIO_DIR global variable
+if [[ -n "$CLAUDE_PROJECT_DIR" ]]; then
+  AUDIO_DIR="$CLAUDE_PROJECT_DIR/.claude/audio"
+else
+  CURRENT_DIR="$PWD"
+  while [[ "$CURRENT_DIR" != "/" ]]; do
+    if [[ -d "$CURRENT_DIR/.claude" ]]; then
+      AUDIO_DIR="$CURRENT_DIR/.claude/audio"
+      break
+    fi
+    CURRENT_DIR=$(dirname "$CURRENT_DIR")
+  done
+  if [[ -z "$AUDIO_DIR" ]]; then
+    AUDIO_DIR="$HOME/.claude/audio"
+  fi
+fi
+
+mkdir -p "$AUDIO_DIR"
+TEMP_FILE="$AUDIO_DIR/tts-$(date +%s).wav"
+
+# @function synthesize_speech
+# @intent Generate speech using best available Soprano mode
+# @why Auto-detect WebUI → API → CLI for optimal performance
+# @param Uses globals: $TEXT, $SOPRANO_PORT, $SOPRANO_DEVICE
+# @returns Creates WAV file at $TEMP_FILE, sets $SYNTH_MODE
+# @exitcode 4=synthesis error
+SYNTH_MODE=""
+
+if check_webui_server; then
+  # Gradio WebUI mode — use Python helper for SSE protocol
+  SYNTH_MODE="webui"
+  python3 "$SCRIPT_DIR/soprano-gradio-synth.py" "$TEXT" "$TEMP_FILE" "$SOPRANO_PORT" 2>/dev/null
+elif check_api_server; then
+  # OpenAI-compatible API mode — direct curl
+  SYNTH_MODE="api"
+  curl -sf "http://127.0.0.1:${SOPRANO_PORT}/v1/audio/speech" \
+    -H "Content-Type: application/json" \
+    -d "$(printf '{"input":"%s"}' "$(echo "$TEXT" | sed 's/"/\\"/g')")" \
+    --output "$TEMP_FILE" 2>/dev/null
+else
+  # CLI fallback — reloads model each call (slowest)
+  SYNTH_MODE="cli"
+  soprano "$TEXT" -o "$TEMP_FILE" -d "$SOPRANO_DEVICE" 2>/dev/null
+fi
+
+if [[ ! -f "$TEMP_FILE" ]] || [[ ! -s "$TEMP_FILE" ]]; then
+  echo "❌ Failed to synthesize speech with Soprano ($SYNTH_MODE mode)"
+  [[ "$SYNTH_MODE" == "webui" ]] && echo "   Try: python3 $SCRIPT_DIR/soprano-gradio-synth.py \"test\" /tmp/test.wav $SOPRANO_PORT"
+  exit 4
+fi
+
+# @function detect_remote_session
+# @intent Auto-detect SSH/RDP sessions and enable audio compression
+# @why Remote desktop audio is choppy without compression
+# @returns Sets AGENTVIBES_RDP_MODE environment variable
+if [[ -z "${AGENTVIBES_RDP_MODE:-}" ]]; then
+  if [[ -n "${SSH_CLIENT:-}" ]] || [[ -n "${SSH_TTY:-}" ]] || [[ "${DISPLAY:-}" =~ ^localhost:.* ]]; then
+    export AGENTVIBES_RDP_MODE=true
+    echo "🌐 Remote session detected - enabling audio compression"
+  fi
+fi
+
+# @function compress_for_remote
+# @intent Compress TTS audio for remote sessions (SSH/RDP)
+# @why Reduces bandwidth and prevents choppy playback
+if [[ "${AGENTVIBES_RDP_MODE:-false}" == "true" ]] && command -v ffmpeg &>/dev/null; then
+  COMPRESSED_FILE="$AUDIO_DIR/tts-compressed-$(date +%s).wav"
+  ffmpeg -i "$TEMP_FILE" -ac 1 -ar 22050 -b:a 64k -y "$COMPRESSED_FILE" 2>/dev/null
+  if [[ -f "$COMPRESSED_FILE" ]]; then
+    rm -f "$TEMP_FILE"
+    TEMP_FILE="$COMPRESSED_FILE"
+  fi
+fi
+
+# @function add_silence_padding
+# @intent Add silence to prevent WSL audio static
+# @why WSL audio subsystem cuts off first ~200ms
+if command -v ffmpeg &>/dev/null; then
+  PADDED_FILE="$AUDIO_DIR/tts-padded-$(date +%s).wav"
+  ffmpeg -f lavfi -i anullsrc=r=44100:cl=stereo:d=0.2 -i "$TEMP_FILE" \
+    -filter_complex "[0:a][1:a]concat=n=2:v=0:a=1[out]" \
+    -map "[out]" -y "$PADDED_FILE" 2>/dev/null
+  if [[ -f "$PADDED_FILE" ]]; then
+    rm -f "$TEMP_FILE"
+    TEMP_FILE="$PADDED_FILE"
+  fi
+fi
+
+# @function apply_audio_effects
+# @intent Apply sox effects and background music via audio-processor.sh
+# @param Uses global: $TEMP_FILE
+# @returns Updates $TEMP_FILE to processed version
+BACKGROUND_MUSIC=""
+if [[ -f "$SCRIPT_DIR/audio-processor.sh" ]]; then
+  PROCESSED_FILE="$AUDIO_DIR/tts-processed-$(date +%s).wav"
+  PROCESSOR_OUTPUT=$("$SCRIPT_DIR/audio-processor.sh" "$TEMP_FILE" "default" "$PROCESSED_FILE" 2>/dev/null) || {
+    PROCESSED_FILE="$TEMP_FILE"
+    PROCESSOR_OUTPUT="$TEMP_FILE|"
+  }
+  PROCESSED_FILE="${PROCESSOR_OUTPUT%%|*}"
+  BACKGROUND_MUSIC="${PROCESSOR_OUTPUT##*|}"
+  if [[ -f "$PROCESSED_FILE" ]] && [[ "$PROCESSED_FILE" != "$TEMP_FILE" ]]; then
+    rm -f "$TEMP_FILE"
+    TEMP_FILE="$PROCESSED_FILE"
+  fi
+fi
+
+# @function play_audio
+# @intent Play generated audio using available player with sequential playback
+# @why Support multiple audio players and prevent overlapping audio
+LOCK_FILE="/tmp/agentvibes-audio.lock"
+
+for i in {1..4}; do
+  if [ ! -f "$LOCK_FILE" ]; then
+    break
+  fi
+  sleep 0.5
+done
+
+if [ -f "$LOCK_FILE" ]; then
+  echo "⏭️  Skipping TTS (previous audio still playing)" >&2
+  exit 0
+fi
+
+touch "$LOCK_FILE"
+
+AUDIO_DIR_PLAY="${TEMP_FILE%/*}"
+WRITE_LOCK_FILE="$AUDIO_DIR_PLAY/$(basename "$TEMP_FILE" .wav).lock"
+touch "$WRITE_LOCK_FILE"
+
+DURATION=$(ffprobe -v error -show_entries format=duration -of default=noprint_wrappers=1:nokey=1 "$TEMP_FILE" 2>/dev/null)
+DURATION=${DURATION%.*}
+DURATION=${DURATION:-1}
+
+if [[ "${AGENTVIBES_TEST_MODE:-false}" != "true" ]] && [[ "${AGENTVIBES_NO_PLAYBACK:-false}" != "true" ]]; then
+  if [[ "$(uname -s)" == "Darwin" ]]; then
+    afplay "$TEMP_FILE" >/dev/null 2>&1 &
+    PLAYER_PID=$!
+  elif [[ -n "${TERMUX_VERSION:-}" ]] || [[ -d "/data/data/com.termux" ]]; then
+    termux-media-player play "$TEMP_FILE" >/dev/null 2>&1 &
+    PLAYER_PID=$!
+  else
+    (paplay "$TEMP_FILE" || mpv "$TEMP_FILE" || aplay "$TEMP_FILE") >/dev/null 2>&1 &
+    PLAYER_PID=$!
+  fi
+fi
+
+(sleep $DURATION; rm -f "$LOCK_FILE" "$WRITE_LOCK_FILE") &
+disown
+
+# @function display_cache_stats
+# @intent Show audio cache statistics with color-coded output
+AUDIO_DIR_PATH=$(get_audio_dir)
+
+BLUE='\033[0;34m'
+YELLOW='\033[1;33m'
+PURPLE='\033[0;35m'
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+ORANGE='\033[0;33m'
+WHITE='\033[1;37m'
+CYAN='\033[0;36m'
+GOLD='\033[38;5;226m'
+NC='\033[0m'
+
+AUTO_CLEAN_THRESHOLD=$(get_auto_clean_threshold)
+INITIAL_SIZE=$(calculate_tts_size_bytes "$AUDIO_DIR_PATH")
+if [[ $INITIAL_SIZE -gt $((AUTO_CLEAN_THRESHOLD * 1048576)) ]]; then
+  DELETED=$(auto_clean_old_files "$AUDIO_DIR_PATH" "$AUTO_CLEAN_THRESHOLD")
+  if [[ $DELETED -gt 0 ]]; then
+    echo -e "${ORANGE}🧹 Auto-cleaned $DELETED old files${NC}"
+  fi
+fi
+
+FILE_COUNT=$(count_tts_files "$AUDIO_DIR_PATH")
+SIZE_BYTES=$(calculate_tts_size_bytes "$AUDIO_DIR_PATH")
+SIZE_HUMAN=$(bytes_to_human "$SIZE_BYTES")
+
+CACHE_COLOR=$GREEN
+if [[ $SIZE_BYTES -gt 3221225472 ]]; then
+  CACHE_COLOR=$RED
+elif [[ $SIZE_BYTES -gt 524288000 ]]; then
+  CACHE_COLOR=$YELLOW
+fi
+
+echo -e "${WHITE}💾 Saved to:${NC} ${CYAN}$TEMP_FILE${NC} ${YELLOW}$FILE_COUNT${NC} ${WHITE}🗄️${NC} ${CACHE_COLOR}$SIZE_HUMAN${NC} ${WHITE}🧹${NC}${GOLD}[${AUTO_CLEAN_THRESHOLD}mb]${NC}"
+
+if [[ -n "$BACKGROUND_MUSIC" ]]; then
+  MUSIC_FILENAME=$(basename "$BACKGROUND_MUSIC")
+  echo -e "${WHITE}🎵 Background music:${NC} ${PURPLE}$MUSIC_FILENAME${NC}"
+fi
+echo -e "${WHITE}🎤 Voice:${NC} ${BLUE}Soprano-1.1-80M${NC} ${WHITE}(Soprano TTS, ${SYNTH_MODE} mode)${NC}"
+
+# Show personality if configured
+PROJECT_ROOT="${PROJECT_ROOT:-$(cd "$SCRIPT_DIR/../.." && pwd)}"
+PERSONALITY=$(cat "$PROJECT_ROOT/.claude/tts-personality.txt" 2>/dev/null || cat "$HOME/.claude/tts-personality.txt" 2>/dev/null || echo "")
+if [[ -n "$PERSONALITY" ]] && [[ "$PERSONALITY" != "none" ]] && [[ "$PERSONALITY" != "normal" ]]; then
+  echo -e "${WHITE}💫 Personality:${NC} ${YELLOW}$PERSONALITY${NC}"
+fi
+
+if [[ -d "$AUDIO_DIR_PATH" ]]; then
+  AUDIO_SIZE=$(du -sm "$AUDIO_DIR_PATH" 2>/dev/null | cut -f1)
+  if [[ -n "$AUDIO_SIZE" ]] && [[ "$AUDIO_SIZE" -gt 100 ]]; then
+    echo -e "\033[0;31m⚠️  Audio cache is ${AUDIO_SIZE}MB - Run: /agent-vibes:cleanup\033[0m"
+  fi
+fi
+
+# Background music status
+if [[ -z "$BACKGROUND_MUSIC" ]]; then
+  BACKGROUND_ENABLED_FILE="$PROJECT_ROOT/.claude/config/background-music-enabled.txt"
+  if [[ -f "$BACKGROUND_ENABLED_FILE" ]] && grep -q "true" "$BACKGROUND_ENABLED_FILE" 2>/dev/null; then
+    echo -e "${WHITE}🎵 Background music:${NC} ${PURPLE}Enabled but not playing (check config)${NC}"
+  else
+    echo -e "${WHITE}🎵 Background music:${NC} ${PURPLE}Disabled${NC}"
+  fi
+fi

package/.claude/hooks/play-tts.sh

@@ -146,6 +146,9 @@ speak_text() {
     piper)
       "$SCRIPT_DIR/play-tts-piper.sh" "$text" "$voice"
       ;;
+    soprano)
+      "$SCRIPT_DIR/play-tts-soprano.sh" "$text" "$voice"
+      ;;
     macos)
       "$SCRIPT_DIR/play-tts-macos.sh" "$text" "$voice"
       ;;
@@ -265,6 +268,9 @@ case "$ACTIVE_PROVIDER" in
   piper)
     exec "$SCRIPT_DIR/play-tts-piper.sh" "$TEXT" "$VOICE_OVERRIDE"
     ;;
+  soprano)
+    exec "$SCRIPT_DIR/play-tts-soprano.sh" "$TEXT" "$VOICE_OVERRIDE"
+    ;;
   macos)
     exec "$SCRIPT_DIR/play-tts-macos.sh" "$TEXT" "$VOICE_OVERRIDE"
     ;;

package/.claude/hooks/provider-manager.sh

@@ -190,6 +190,13 @@ migrate_voice_to_provider() {
   # Default voices by provider
   local piper_default="en_US-lessac-medium"
   local macos_default="Samantha"
+  local soprano_default="soprano-default"  # Single voice — no selection needed
+
+  # Soprano has a single voice, so migration is straightforward
+  if [[ "$target_provider" == "soprano" ]]; then
+    echo "$soprano_default"
+    return 0
+  fi
 
   # If no current voice, return default for target provider
   if [[ -z "$current_voice" ]]; then
@@ -201,6 +208,16 @@ migrate_voice_to_provider() {
     return 0
   fi
 
+  # If migrating FROM Soprano, return default for target provider
+  if [[ "$current_voice" == "soprano-default" ]]; then
+    case "$target_provider" in
+      piper) echo "$piper_default" ;;
+      macos) echo "$macos_default" ;;
+      *) echo "$piper_default" ;;
+    esac
+    return 0
+  fi
+
   # Convert to lowercase for case-insensitive comparison (portable)
   local current_voice_lower
   current_voice_lower=$(echo "$current_voice" | tr '[:upper:]' '[:lower:]')

package/.claude/hooks/soprano-gradio-synth.py

@@ -0,0 +1,139 @@
+#!/usr/bin/env python3
+#
+# File: .claude/hooks/soprano-gradio-synth.py
+#
+# AgentVibes - Finally, your AI Agents can Talk Back!
+# Website: https://agentvibes.org
+# Repository: https://github.com/paulpreibisch/AgentVibes
+#
+# Co-created by Paul Preibisch with Claude AI
+# Copyright (c) 2025 Paul Preibisch
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+# ---
+#
+# Soprano Gradio WebUI synthesizer helper.
+# Calls the Soprano WebUI's Gradio API and saves the result as a WAV file.
+# Uses only Python stdlib (json, sys, urllib) — no extra dependencies.
+#
+# Usage: python3 soprano-gradio-synth.py "text to speak" output.wav [port]
+#
+"""
+Soprano Gradio WebUI synthesizer helper for AgentVibes.
+
+Calls the Soprano WebUI's Gradio API via the Server-Sent Events (SSE) protocol
+and downloads the generated audio as a WAV file.
+
+Flow:
+  1. Submit generation request → get event_id
+  2. Poll SSE stream for audio file URL
+  3. Download WAV file to output path
+
+See: https://github.com/ekwek1/soprano
+"""
+import json
+import sys
+import urllib.request
+import urllib.error
+
+
+def synth(text: str, output_path: str, port: int = 7860) -> None:
+    base = f"http://127.0.0.1:{port}"
+
+    # Step 1: Submit generation request
+    # Args: text, temperature, top_p, repetition_penalty, chunk_size, streaming
+    payload = json.dumps({
+        "data": [text, 0.0, 0.95, 1.2, 1, False]
+    }).encode()
+
+    event_id = submit_request(base, payload)
+
+    # Step 2: Poll SSE stream for audio file URL
+    audio_url = poll_for_result(base, event_id)
+
+    # Step 3: Download the audio file
+    download_file(audio_url, output_path)
+
+
+def submit_request(base: str, payload: bytes) -> str:
+    """Submit generation request to Gradio API, return event_id."""
+    for api_base in ["/gradio_api/call", "/call"]:
+        url = f"{base}{api_base}/generate_speech"
+        req = urllib.request.Request(
+            url,
+            data=payload,
+            headers={"Content-Type": "application/json"},
+        )
+        try:
+            with urllib.request.urlopen(req, timeout=30) as resp:
+                return json.loads(resp.read())["event_id"]
+        except urllib.error.URLError:
+            continue
+
+    raise RuntimeError("Could not reach Soprano WebUI API")
+
+
+def poll_for_result(base: str, event_id: str) -> str:
+    """Poll SSE endpoint until audio file URL is returned."""
+    for api_base in ["/gradio_api/call", "/call"]:
+        url = f"{base}{api_base}/generate_speech/{event_id}"
+        req = urllib.request.Request(url)
+        try:
+            with urllib.request.urlopen(req, timeout=120) as resp:
+                for raw_line in resp:
+                    line = raw_line.decode("utf-8", errors="replace").strip()
+                    if not line.startswith("data: "):
+                        continue
+                    try:
+                        parsed = json.loads(line[6:])
+                    except json.JSONDecodeError:
+                        continue
+                    # Response format: [{"path": "...", "url": "...", ...}, "status string"]
+                    if isinstance(parsed, list) and len(parsed) >= 1:
+                        audio = parsed[0]
+                        if isinstance(audio, dict) and "url" in audio:
+                            return audio["url"]
+            break
+        except urllib.error.URLError:
+            continue
+
+    raise RuntimeError("No audio URL in Soprano response")
+
+
+def download_file(url: str, output_path: str) -> None:
+    """Download audio file from Gradio file server."""
+    req = urllib.request.Request(url)
+    with urllib.request.urlopen(req, timeout=30) as resp:
+        with open(output_path, "wb") as f:
+            while True:
+                chunk = resp.read(8192)
+                if not chunk:
+                    break
+                f.write(chunk)
+
+
+if __name__ == "__main__":
+    if len(sys.argv) < 3:
+        print(f"Usage: {sys.argv[0]} \"text\" output.wav [port]", file=sys.stderr)
+        sys.exit(1)
+
+    text = sys.argv[1]
+    output = sys.argv[2]
+    port = int(sys.argv[3]) if len(sys.argv) > 3 else 7860
+
+    try:
+        synth(text, output, port)
+    except Exception as e:
+        print(f"Error: {e}", file=sys.stderr)
+        sys.exit(1)

package/.claude/piper-voices-dir.txt

@@ -1 +1 @@
-/home/fire/.claude/piper-voices
+/home/administrator/.claude/piper-voices

package/.mcp.json

@@ -1,34 +1,12 @@
 {
   "mcpServers": {
-    "sonarqube": {
-      "command": "docker",
+    "agentvibes": {
+      "command": "npx",
       "args": [
-        "run",
-        "-i",
-        "--name",
-        "sonarqube-mcp-server-agentvibes",
-        "--rm",
-        "--dns",
-        "8.8.8.8",
-        "--dns",
-        "8.8.4.4",
-        "-e",
-        "SONARQUBE_TOKEN",
-        "-e",
-        "SONARQUBE_ORG",
-        "-e",
-        "STORAGE_PATH",
-        "mcp/sonarqube"
-      ],
-      "env": {
-        "SONARQUBE_TOKEN": "${SONARQUBE_TOKEN}",
-        "SONARQUBE_ORG": "${SONARQUBE_ORG}",
-        "STORAGE_PATH": "${STORAGE_PATH}"
-      }
-    },
-    "pieces": {
-      "type": "sse",
-      "url": "http://localhost:39300/model_context_protocol/2024-11-05/sse"
+        "-y",
+        "--package=agentvibes",
+        "agentvibes-mcp-server"
+      ]
     }
   }
 }