Skip to content

World's first open-source, agentic video production system. 12 production pipelines, 100+ tools, 700+ agent skill and production-knowledge files. Turn your AI coding assistant into a full video production studio.

fix: accept voice_id alias and derive language_code in google_tts (#267)

ytmknd4fafc6bc1184
diff --git a/tests/tools/test_google_tts_voice_id_alias.py b/tests/tools/test_google_tts_voice_id_alias.py
new file mode 100644
index 00000000..b7bb2f34
--- /dev/null
+++ b/tests/tools/test_google_tts_voice_id_alias.py
@@ -0,0 +1,163 @@
+"""Regression tests for google_tts voice_id alias + language_code derivation.
+
+Covers the selector-compatibility fix: tts_selector passes its
+provider-agnostic 'voice_id' through unchanged, so google_tts must accept
+it as an alias for 'voice' and derive languageCode from the voice name's
+BCP-47 locale prefix. No live API calls: requests.post is monkeypatched
+and the captured payload is asserted directly.
+"""
+
+import base64
+import sys
+from pathlib import Path
+
+import pytest
+
+PROJECT_ROOT = Path(__file__).resolve().parent.parent.parent
+sys.path.insert(0, str(PROJECT_ROOT))
+
+from tools.audio.google_tts import GoogleTTS
+
+
+FAKE_AUDIO = b"fake-mp3-bytes"
+
+
+class _FakeResponse:
+ status_code = 200
+
+ def json(self):
+ return {"audioContent": base64.b64encode(FAKE_AUDIO).decode()}
+
+ def raise_for_status(self):
+ return None
+
+
+def captured_synthesize(monkeypatch):
+ """Route requests.post to a fake and capture the synthesize call."""
+ calls = []
+
+ def fake_post(url, headers=None, params=None, json=None, timeout=None):
+ calls.append({"url": url, "params": params, "payload": json})
+ return _FakeResponse()
+
+ monkeypatch.setenv("GOOGLE_API_KEY", "fake-key")
+ monkeypatch.delenv("GEMINI_API_KEY", raising=False)
+ monkeypatch.setattr("requests.post", fake_post)
+ return calls
+
+
+def _execute(tmp_path, **inputs):
+ inputs.setdefault("text", "こんにちは")
+ inputs.setdefault("output_path", str(tmp_path / "out.mp3"))
+ return GoogleTTS().execute(inputs)
+
+
+# ---- voice_id alias → request payload ----
+
+
+class TestVoiceIdAlias:
+ def test_voice_id_sets_voice_name_and_derived_language(
+ self, tmp_path, captured_synthesize
+ ):
+ result = _execute(tmp_path, voice_id="ja-JP-Chirp3-HD-Kore")
+
+ assert result.success, result.error
+ voice_payload = captured_synthesize[0]["payload"]["voice"]
+ assert voice_payload["name"] == "ja-JP-Chirp3-HD-Kore"
+ assert voice_payload["languageCode"] == "ja-JP"
+ assert result.data["voice"] == "ja-JP-Chirp3-HD-Kore"
+ assert result.data["language_code"] == "ja-JP"
+
+ def test_explicit_voice_wins_over_voice_id(self, tmp_path, captured_synthesize):
+ result = _execute(
+ tmp_path,
+ voice="es-ES-Neural2-A",
+ voice_id="ja-JP-Chirp3-HD-Kore",
+ )
+
+ assert result.success, result.error
+ voice_payload = captured_synthesize[0]["payload"]["voice"]
+ assert voice_payload["name"] == "es-ES-Neural2-A"
+ assert voice_payload["languageCode"] == "es-ES"
+
+ def test_explicit_language_code_wins_over_derived(
+ self, tmp_path, captured_synthesize
+ ):
+ result = _execute(
+ tmp_path,
+ voice_id="ja-JP-Chirp3-HD-Kore",
+ language_code="en-US",
+ )
+
+ assert result.success, result.error
+ voice_payload = captured_synthesize[0]["payload"]["voice"]
+ assert voice_payload["name"] == "ja-JP-Chirp3-HD-Kore"
+ assert voice_payload["languageCode"] == "en-US"
+
+ def test_default_voice_when_neither_field_set(self, tmp_path, captured_synthesize):
+ result = _execute(tmp_path)
+
+ assert result.success, result.error
+ voice_payload = captured_synthesize[0]["payload"]["voice"]
+ assert voice_payload["name"] == "en-US-Chirp3-HD-Orus"
+ assert voice_payload["languageCode"] == "en-US"
+
+ def test_chirp_voice_id_routes_to_beta_endpoint(
+ self, tmp_path, captured_synthesize
+ ):
+ # Endpoint selection must see the resolved alias, not just 'voice'.
+ _execute(tmp_path, voice_id="ja-JP-Chirp3-HD-Kore")
+ assert "/v1beta1/" in captured_synthesize[0]["url"]
+
+ _execute(tmp_path, voice_id="ja-JP-Neural2-B")
+ assert "/v1/" in captured_synthesize[1]["url"]
+
+
+# ---- locale derivation ----
+
+
+class TestLocaleDerivation:
+ def test_two_letter_prefix(self):
+ assert GoogleTTS._locale_from_voice("ja-JP-Chirp3-HD-Kore") == "ja-JP"
+
+ def test_three_letter_prefix(self):
+ assert GoogleTTS._locale_from_voice("fil-PH-Standard-A") == "fil-PH"
+
+ def test_unparseable_name_falls_back_to_en_us(self):
+ assert GoogleTTS._locale_from_voice("Kore") == "en-US"
+
+
+# ---- cost & idempotency (no API call) ----
+
+
+class TestCostAndIdempotency:
+ def test_voice_id_selects_pricing_tier(self):
+ tool = GoogleTTS()
+ text = "a" * 1000
+ chirp_cost = tool.estimate_cost(
+ {"text": text, "voice_id": "ja-JP-Chirp3-HD-Kore"}
+ )
+ standard_cost = tool.estimate_cost(
+ {"text": text, "voice_id": "ja-JP-Standard-B"}
+ )
+ # Chirp 3 HD ($30/1M chars) vs Standard ($4/1M chars)
+ assert chirp_cost == pytest.approx(0.03)
+ assert standard_cost == pytest.approx(0.004)
+
+ def test_voice_id_matches_equivalent_voice_pricing(self):
+ tool = GoogleTTS()
+ inputs_alias = {"text": "hello", "voice_id": "en-US-Studio-O"}
+ inputs_direct = {"text": "hello", "voice": "en-US-Studio-O"}
+ assert tool.estimate_cost(inputs_alias) == tool.estimate_cost(inputs_direct)
+
+ def test_voice_id_changes_idempotency_key(self):
+ tool = GoogleTTS()
+ assert "voice_id" in tool.idempotency_key_fields
+ key_kore = tool.idempotency_key(
+ {"text": "hello", "voice_id": "ja-JP-Chirp3-HD-Kore"}
+ )
+ key_orus = tool.idempotency_key(
+ {"text": "hello", "voice_id": "en-US-Chirp3-HD-Orus"}
+ )
+ assert key_kore != key_orus
diff --git a/tools/audio/google_tts.py b/tools/audio/google_tts.py
index f47dd4d6..c48fbe42 100644
--- a/tools/audio/google_tts.py
+++ b/tools/audio/google_tts.py
@@ -93,10 +93,13 @@ class GoogleTTS(BaseTool):
"default": "en-US-Chirp3-HD-Orus",
"description": "Voice name. Default tier is Chirp 3 HD (2024, most natural). Examples: en-US-Chirp3-HD-Orus (male, rich/cinematic), en-US-Chirp3-HD-Aoede (female, warm). Legacy tiers: en-US-Studio-O, en-US-Neural2-D, en-US-Journey-D.",
},
+ "voice_id": {
+ "type": "string",
+ "description": "Alias for 'voice' — accepted so tts_selector's provider-agnostic voice_id passes through. 'voice' wins when both are set.",
+ },
"language_code": {
"type": "string",
- "default": "en-US",
- "description": "BCP-47 language code (e.g. en-US, es-ES, ja-JP, fr-FR)",
+ "description": "BCP-47 language code (e.g. en-US, es-ES, ja-JP, fr-FR). Defaults to the locale prefix of the voice name (e.g. ja-JP for ja-JP-Chirp3-HD-Kore), falling back to en-US.",
},
"speaking_rate": {
"type": "number",
@@ -132,6 +135,7 @@ class GoogleTTS(BaseTool):
"text",
"input_type",
"voice",
+ "voice_id",
"language_code",
"speaking_rate",
"pitch",
@@ -169,10 +173,18 @@ class GoogleTTS(BaseTool):
"""Check if voice requires the v1beta1 endpoint."""
return any(prefix in voice for prefix in self._BETA_VOICE_PREFIXES)
+ @staticmethod
+ def _locale_from_voice(voice_name: str) -> str:
+ """Derive the BCP-47 locale from a voice name like ja-JP-Chirp3-HD-Kore."""
+ import re
+
+ match = re.match(r"^([a-z]{2,3}-[A-Z]{2})-", voice_name)
+ return match.group(1) if match else "en-US"
+
def estimate_cost(self, inputs: dict[str, Any]) -> float:
text = inputs.get("text", "")
char_count = len(text)
- voice = inputs.get("voice", "en-US-Chirp3-HD-Orus")
+ voice = inputs.get("voice") or inputs.get("voice_id") or "en-US-Chirp3-HD-Orus"
# Pricing per million characters (approximate)
if "Chirp3-HD" in voice:
rate_per_char = 0.000030 # $30/1M chars
@@ -228,8 +240,8 @@ class GoogleTTS(BaseTool):
text = inputs["text"]
input_type = inputs.get("input_type", "text")
- voice_name = inputs.get("voice", "en-US-Chirp3-HD-Orus")
- language_code = inputs.get("language_code", "en-US")
+ voice_name = inputs.get("voice") or inputs.get("voice_id") or "en-US-Chirp3-HD-Orus"
+ language_code = inputs.get("language_code") or self._locale_from_voice(voice_name)
speaking_rate = inputs.get("speaking_rate", 1.0)
pitch = inputs.get("pitch", 0.0)
audio_encoding = inputs.get("audio_encoding", "MP3")