agentvibes 5.13.0 → 5.14.0

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

@@ -47,8 +47,13 @@ if [[ -z "$PYTHON_BIN" ]]; then
   echo "   or set AGENTVIBES_PYTHON=/path/to/python.exe if it's installed elsewhere." >&2
   exit 2
 fi
-# Check kokoro is installed (use find_spec to avoid slow torch import)
-if ! "$PYTHON_BIN" -c "import importlib.util; exit(0 if importlib.util.find_spec('kokoro') else 1)" 2>/dev/null; then
+# Check kokoro is installed (use find_spec to avoid slow torch import).
+# Skipped in AGENTVIBES_TEST_MODE: the hermetic sentinel tests emit a fake
+# AV_OUTPUT below without real synthesis, so kokoro need not be installed on the
+# runner (mirrors the TEST_MODE guards in play-tts-piper.sh; CI has Piper but not
+# the heavy Kokoro deps).
+if [[ "${AGENTVIBES_TEST_MODE:-false}" != "true" ]] && \
+   ! "$PYTHON_BIN" -c "import importlib.util; exit(0 if importlib.util.find_spec('kokoro') else 1)" 2>/dev/null; then
   echo "❌ Kokoro TTS module not installed for: $PYTHON_BIN" >&2
   echo "   Install with: ${SCRIPT_DIR}/kokoro-installer.sh" >&2
   echo "   Or manually:  \"$PYTHON_BIN\" -m pip install kokoro soundfile numpy" >&2

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

@@ -15,6 +15,7 @@
 set -euo pipefail
 
 TEXT="${1:-}"
+_VOICE_ARG="${2:-}"                 # raw voice as passed (empty on normal hook TTS)
 VOICE="${2:-en_US-lessac-medium}"
 AGENT_NAME="${3:-default}"
 AGENT_PROFILE_FILE="${4:-}"
@@ -305,10 +306,39 @@ fi
 
 # Read the TTS provider the RECEIVER should use to generate audio.
 # This is separate from the sender's own provider (which is "ssh-remote").
-# Check receiver-provider.txt first, then fall back to "piper".
+# Priority: one-off caller override (a voice-picker PREVIEW forces the previewed
+# voice's own engine for THIS send only) → project receiver-provider.txt →
+# home receiver-provider.txt → "piper". The override intentionally does NOT
+# write any config: a preview must preview, never change the persisted setting.
 PROVIDER=""
+if [[ -n "${AGENTVIBES_RECEIVER_PROVIDER_OVERRIDE:-}" ]]; then
+  PROVIDER="$AGENTVIBES_RECEIVER_PROVIDER_OVERRIDE"
+fi
+# Voice->engine coupling (mirrors the guard in play-tts.ps1): a Kokoro-format
+# voice id (strict lang/gender prefix, e.g. af_bella) or a Piper voice (locale+
+# hyphen like en_US-lessac-medium, or an ::Speaker suffix) can ONLY be
+# synthesised by that engine — so the voice itself decides the receiver engine,
+# overriding a stale/mislabeled provider default (e.g. an old console that always
+# sends windows-sapi). Names that don't encode an engine (SAPI/macOS names) fall
+# through to the configured provider below.
+#
+# GATED on an EXPLICIT voice arg ($2): normal hook TTS passes no voice, and VOICE
+# then defaults to a piper id — coupling on that default would wrongly force
+# `piper` on EVERY normal utterance and shadow receiver-provider.txt (a kokoro/
+# SAPI receiver would silently regress to piper). Only a caller that passed a
+# real voice (every voice-picker preview does) gets voice->engine coupling.
+if [[ -z "$PROVIDER" && -n "$_VOICE_ARG" ]]; then
+  _v_nospk="${VOICE%%::*}"   # strip any ::Speaker-ID suffix before matching
+  # Kokoro ids use a fixed lang/gender prefix set (mirrors play-tts.sh) — a loose
+  # `^[a-z]{2}_...` would misclassify a custom piper model like `my_voice`.
+  if [[ "$_v_nospk" =~ ^(af|am|bf|bm|jf|jm|kf|km|zf|zm|ff|fm|hf|hm|if|im|pf|pm|ef|em|nf|nm)_[a-z0-9_]+$ ]]; then
+    PROVIDER="kokoro"
+  elif [[ "$VOICE" == *"::"* ]] || [[ "$_v_nospk" =~ ^[a-z]{2}_[A-Za-z]{2}-. ]]; then
+    PROVIDER="piper"
+  fi
+fi
 RECEIVER_PROVIDER_FILE="$PROJECT_ROOT/.agentvibes/config/receiver-provider.txt"
-if [[ -f "$RECEIVER_PROVIDER_FILE" ]]; then
+if [[ -z "$PROVIDER" && -f "$RECEIVER_PROVIDER_FILE" ]]; then
   PROVIDER=$(cat "$RECEIVER_PROVIDER_FILE" 2>/dev/null || true)
 fi
 # Also check home-level config
@@ -392,7 +422,12 @@ fi
 # SECURITY: Use jq if available for safe JSON construction, else manual escaping
 build_json_payload() {
   if command -v jq &>/dev/null; then
-    jq -n \
+    # On Windows git-bash, native jq.exe triggers MSYS2 path-conversion of any
+    # POSIX-looking --arg value (projectPath "/c/proj" becomes "C:/proj", music
+    # file paths likewise), silently corrupting the JSON string the receiver
+    # reads. Excluding all args from conversion passes every --arg literally.
+    # Both vars are no-ops on Linux/macOS.
+    MSYS2_ARG_CONV_EXCL='*' MSYS_NO_PATHCONV=1 jq -n \
       --arg text "$TEXT" \
       --arg voice "$VOICE" \
       --arg effects "$SOX_EFFECTS" \

package/.claude/hooks/requirements.txt

@@ -1,6 +1,6 @@
-# AgentVibes TTS Hooks Requirements
-# Install with: pip install -r requirements.txt
-
-# Translation support for multi-language TTS and learning mode
-deep-translator>=1.11.4
-langdetect>=1.0.9
+# AgentVibes TTS Hooks Requirements
+# Install with: pip install -r requirements.txt
+
+# Translation support for multi-language TTS and learning mode
+deep-translator>=1.11.4
+langdetect>=1.0.9

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

@@ -1,139 +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)
+#!/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)