0.1.7

* support noise isolation to config verb

* wip

* wip

* wip

* wip

* add vendor to turnTaking

jambonz-app-json-validation.js

@@ -4,6 +4,27 @@ const _specData = require('./specs');
 const specs = new Map();
 for (const key in _specData) { specs.set(key, _specData[key]); }
 
+/* verb synonyms and shortcuts: maps alias verb names to their canonical form,
+   optionally injecting properties (e.g. vendor) into the verb data */
+const verbTransforms = new Map([
+  ['stream', {verb: 'listen'}],
+  ['s2s', {verb: 'llm'}],
+  ['openai_s2s', {verb: 'llm', properties: {vendor: 'openai'}}],
+  ['microsoft_s2s', {verb: 'llm', properties: {vendor: 'microsoft'}}],
+  ['google_s2s', {verb: 'llm', properties: {vendor: 'google'}}],
+  ['elevenlabs_s2s', {verb: 'llm', properties: {vendor: 'elevenlabs'}}],
+  ['deepgram_s2s', {verb: 'llm', properties: {vendor: 'deepgram'}}],
+  ['voiceagent_s2s', {verb: 'llm', properties: {vendor: 'voiceagent'}}],
+  ['ultravox_s2s', {verb: 'llm', properties: {vendor: 'ultravox'}}],
+]);
+
+function applyVerbTransform(name, data) {
+  const transform = verbTransforms.get(name);
+  if (!transform) return {name, data};
+  const newData = transform.properties ? {...transform.properties, ...data} : data;
+  return {name: transform.verb, data: newData};
+}
+
 function normalizeJambones(logger, obj) {
   if (!Array.isArray(obj)) {
     throw new Error('malformed jambonz payload: must be array');
@@ -13,18 +34,22 @@ function normalizeJambones(logger, obj) {
     if (typeof tdata !== 'object') throw new Error('malformed jambonz payload: must be array of objects');
     if ('verb' in tdata) {
       // {verb: 'say', text: 'foo..bar'..}
-      const name = tdata.verb;
       const o = {};
       Object.keys(tdata)
         .filter((k) => k !== 'verb')
         .forEach((k) => o[k] = tdata[k]);
+      const {name, data} = applyVerbTransform(tdata.verb, o);
       const o2 = {};
-      o2[name] = o;
+      o2[name] = data;
       document.push(o2);
     }
     else if (Object.keys(tdata).length === 1) {
       // {'say': {..}}
-      document.push(tdata);
+      const key = Object.keys(tdata)[0];
+      const {name, data} = applyVerbTransform(key, tdata[key]);
+      const o2 = {};
+      o2[name] = data;
+      document.push(o2);
     }
     else {
       logger.info(tdata, 'malformed jambonz payload: missing verb property');

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@jambonz/verb-specifications",
-  "version": "0.0.101",
+  "version": "0.1.7",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "@jambonz/verb-specifications",
-      "version": "0.0.101",
+      "version": "0.1.7",
       "license": "MIT",
       "dependencies": {
         "debug": "^4.3.4",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jambonz/verb-specifications",
-  "version": "0.0.101",
+  "version": "0.1.7",
   "description": "Jambonz Verb Specification Utilities",
   "main": "index.js",
   "scripts": {

specs.json

@@ -1,4 +1,13 @@
 {
+  "alert" : {
+    "properties": {
+      "id": "string",
+      "message": "string"
+    },
+    "required": [
+      "message"
+    ]
+  },
   "answer": {
     "properties": {
       "id": "string"
@@ -57,6 +66,7 @@
       "amd": "#amd",
       "fillerNoise": "#fillerNoise",
       "notifyEvents": "boolean",
+      "notifySttLatency": "boolean",
       "reset": "string|array",
       "onHoldMusic": "string",
       "actionHookDelayAction": "#actionHookDelayAction",
@@ -64,7 +74,12 @@
       "boostAudioSignal": "number|string",
       "vad":"#vad",
       "referHook": "object|string",
-      "earlyMedia": "boolean"
+      "earlyMedia": "boolean",
+      "autoStreamTts": "boolean",
+      "disableTtsCache": "boolean",
+      "trackTtsPlayout": "boolean",
+      "noiseIsolation": "#noiseIsolation",
+      "turnTaking": "#turnTaking"
     },
     "required": []
   },
@@ -213,6 +228,7 @@
     "properties": {
       "id": "string",
       "text": "string|array",
+      "instructions": "string",
       "stream": "boolean",
       "loop": "number|string",
       "synthesizer": "#synthesizer",
@@ -266,7 +282,9 @@
       "statusEvents": "array",
       "statusHook": "object|string",
       "enterHook": "object|string",
-      "record": "#record"
+      "record": "#record",
+      "listen": "#listen",
+      "distributeDtmf": "boolean"
     },
     "required": [
       "name"
@@ -298,7 +316,8 @@
       "transcribe": "#transcribe",
       "amd": "#amd",
       "dub": ["#dub"],
-      "tag": "object"
+      "tag": "object",
+      "forwardPAI": "boolean"
     },
     "required": [
       "target"
@@ -314,7 +333,7 @@
       "region": "string",
       "model": {
         "type": "string",
-        "enum": ["es", "cx"]
+        "enum": ["es", "cx", "ces"]
       },
       "lang": "string",
       "actionHook": "object|string",
@@ -409,33 +428,6 @@
       "url"
     ]
   },
-  "stream": {
-    "properties": {
-      "id": "string",
-      "actionHook": "object|string",
-      "auth": "#auth",
-      "finishOnKey": "string",
-      "maxLength": "number",
-      "metadata": "object",
-      "mixType": {
-        "type": "string",
-        "enum": ["mono", "stereo", "mixed"]
-      },
-      "passDtmf": "boolean",
-      "playBeep": "boolean",
-      "disableBidirectionalAudio": "boolean",
-      "bidirectionalAudio": "#bidirectionalAudio",
-      "sampleRate": "number",
-      "timeout": "number",
-      "transcribe": "#transcribe",
-      "url": "string",
-      "wsAuth": "#auth",
-      "earlyMedia": "boolean"
-    },
-    "required": [
-      "url"
-    ]
-  },
   "llm": {
     "properties": {
       "id": "string",
@@ -443,6 +435,7 @@
       "model": "string",
       "auth": "object",
       "connectOptions": "object",
+      "mcpServers": ["#mcpServer"],
       "actionHook": "object|string",
       "eventHook": "object|string",
       "toolHook": "object|string",
@@ -451,11 +444,19 @@
     },
     "required": [
       "vendor",
-      "model",
-      "auth",
       "llmOptions"
     ]
   },
+  "mcpServer": {
+    "properties": {
+      "url": "string",
+      "auth": "object",
+      "roots": ["#root"]
+    },
+    "required": [
+      "url"
+    ]
+  },
   "message": {
     "properties": {
       "id": "string",
@@ -510,6 +511,11 @@
         "type": "string",
         "enum": ["startCallRecording", "stopCallRecording", "pauseCallRecording", "resumeCallRecording"]
       },
+      "type" : {
+        "type" : "string",
+        "enum" : ["cloud", "siprec"]
+
+      },
       "recordingID": "string",
       "siprecServerURL": "string|array",
       "headers": "object"
@@ -521,7 +527,8 @@
   "redirect": {
     "properties": {
       "id": "string",
-      "actionHook": "object|string"
+      "actionHook": "object|string",
+      "statusHook": "object|string"
     },
     "required": [
       "actionHook"
@@ -577,7 +584,6 @@
       "channel": "number"
     },
     "required": [
-      "transcriptionHook"
     ]
   },
   "target": {
@@ -658,6 +664,7 @@
       "fallbackLabel": "string",
       "fallbackLanguage": "string",
       "vad": "#vad",
+      "autogeneratePrompt": "boolean",
       "hints": "array",
       "hintsBoost": "number",
       "altLanguages": "array",
@@ -734,7 +741,10 @@
       "customOptions": "#customOptions",
       "verbioOptions": "#verbioOptions",
       "speechmaticsOptions": "#speechmaticsOptions",
-      "openaiOptions": "#openaiOptions"
+      "openaiOptions": "#openaiOptions",
+      "houndifyOptions": "#houndifyOptions",
+      "gladiaOptions": "object",
+      "elevenlabsOptions": "#elevenlabsOptions"
     },
     "required": [
       "vendor"
@@ -777,6 +787,7 @@
     "properties": {
       "speechSegmentationSilenceTimeoutMs": "number",
       "postProcessing" : "string",
+      "audioLogging" : "boolean",
       "languageIdMode": {
         "type": "string",
         "enum": [
@@ -812,6 +823,62 @@
       "transcriptNormalization": "array"
     }
   },
+  "houndifyOptions": {
+    "properties": {
+      "requestInfo": "object",
+      "sampleRate": "number",
+      "latitude": "number",
+      "longitude": "number",
+      "city": "string",
+      "state": "string",
+      "country": "string",
+      "timeZone": "string",
+      "domain": "string",
+      "audioEndpoint": "string",
+      "maxSilenceSeconds": "number",
+      "maxSilenceAfterFullQuerySeconds": "number",
+      "maxSilenceAfterPartialQuerySeconds": "number",
+      "vadSensitivity": "number",
+      "vadTimeout": "number",
+      "vadMode": "string",
+      "vadVoiceMs": "number",
+      "vadSilenceMs": "number",
+      "vadDebug": "boolean",
+      "audioFormat": "string",
+      "enableNoiseReduction": "boolean",
+      "enableProfanityFilter": "boolean",
+      "enablePunctuation": "boolean",
+      "enableCapitalization": "boolean",
+      "confidenceThreshold": "number",
+      "enableDisfluencyFilter": "boolean",
+      "maxResults": "number",
+      "enableWordTimestamps": "boolean",
+      "maxAlternatives": "number",
+      "partialTranscriptInterval": "number",
+      "sessionTimeout": "number",
+      "connectionTimeout": "number",
+      "customVocabulary": "array",
+      "languageModel": "string",
+      "audioQueryAbsoluteTimeout": "number"
+    }
+  },
+  "elevenlabsOptions": {
+    "properties": {
+      "includeTimestamps": "boolean",
+      "commitStrategy": {
+        "type": "string",
+        "enum": [
+          "manual",
+          "vad"
+        ]
+      },
+      "vadSilenceThresholdSecs": "number",
+      "vadThreshold": "number",
+      "minSpeechDurationMs": "number",
+      "minSilenceDurationMs": "number",
+      "enableLogging": "boolean"
+    }
+  },
   "cobaltOptions": {
     "properties": {
       "serverUri": "string",
@@ -891,7 +958,12 @@
       "shortUtterance": "boolean",
       "vadTurnoff": "number",
       "tag": "string",
-      "fillerWords" : "boolean"
+      "fillerWords" : "boolean",
+      "eotThreshold": "number",
+      "eotTimeoutMs": "number",
+      "mipOptOut": "boolean",
+      "entityPrompt": "string",
+      "eagerEotThreshold":"number"
     }
   },
   "sonioxOptions": {
@@ -992,6 +1064,7 @@
       "additional_vocab": "array",
       "diarization": "string",
       "speaker_diarization_config": "#sm_speakerDiarizationConfig",
+      "conversation_config": "#sm_conversationConfig",
       "enable_partials": "boolean",
       "max_delay": "number",
       "max_delay_mode": {
@@ -1019,6 +1092,13 @@
     "required": [
     ]
   },
+  "sm_conversationConfig": {
+    "properties": {
+      "end_of_utterance_silence_trigger": "number"
+    },
+    "required": [
+    ]
+  },
   "sm_puctuationOverrides": {
     "properties": {
       "permitted_marks": "array",
@@ -1125,7 +1205,25 @@
   },
   "assemblyAiOptions": {
     "properties": {
-      "apiKey": "string"
+      "apiKey": "string",
+      "serviceVersion": {
+        "type": "string",
+        "enum": [
+          "v2",
+          "v3"
+        ]
+      },
+      "speechModel": "string",
+      "formatTurns": "boolean",
+      "endOfTurnConfidenceThreshold": "number",
+      "minEndOfTurnSilenceWhenConfident": "number",
+      "maxTurnSilence": "number",
+      "minTurnSilence": "number",
+      "keyterms": "array",
+      "prompt": "string",
+      "languageDetection": "boolean",
+      "vadThreshold": "number",
+      "inactivityTimeout": "number"
     }
   },
   "resource": {
@@ -1206,10 +1304,18 @@
       "voiceMs": "number",
       "silenceMs": "number",
       "strategy": "string",
-      "mode": "number"
+      "mode": "number",
+      "vendor": {
+        "type": "string",
+        "enum": [
+          "webrtc",
+          "silero"
+        ]
+      },
+      "threshold": "number",
+      "speechPadMs": "number"
     },
     "required": [
-      "enable"
     ]
   },
   "amd": {
@@ -1248,5 +1354,57 @@
       "streaming": "boolean",
       "sampleRate": "number"
     }
+  },
+  "pipeline": {
+    "properties": {
+      "id": "string",
+      "stt": "#recognizer",
+      "tts": "#synthesizer",
+      "llm": "#llm",
+      "turnDetection": "string|object",
+      "bargeIn": "#bargeInPipeline",
+      "actionHook": "object|string",
+      "eventHook": "object|string",
+      "toolHook": "object|string",
+      "greeting": "boolean",
+      "earlyGeneration": "boolean",
+      "noiseIsolation": "string|#noiseIsolationPipeline"
+    },
+    "required": [
+      "llm"
+    ]
+  },
+  "bargeInPipeline": {
+    "properties": {
+      "enable": "boolean",
+      "minSpeechDuration": "number",
+      "sticky": "boolean"
+    }
+  },
+  "noiseIsolationPipeline": {
+    "properties": {
+      "mode": "string",
+      "level": "number",
+      "direction": {
+        "enum": ["read", "write"]
+      },
+      "model": "string"
+    }
+  },
+  "noiseIsolation" : {
+    "properties": {
+      "enable": "boolean",
+      "vendor": "string",
+      "level": "number",
+      "model": "string"
+    }
+  },
+  "turnTaking": {
+    "properties": {
+      "enable": "boolean",
+      "vendor": "string",
+      "threshold": "number",
+      "model": "string"
+    }
   }
 }

test/jambonz-verb-test.js

@@ -1,6 +1,6 @@
 const test = require('tape');
 const logger = require('pino')({level: process.env.JAMBONES_LOGLEVEL || 'error'});
-const { validate } = require('..');
+const { validate, normalizeJambones } = require('..');
 
 test("validate correct verbs", async (t) => {
 
@@ -72,12 +72,18 @@ test("validate correct verbs", async (t) => {
     },
     {
       "verb": "config",
+      "notifySttLatency": true,
       "recognizer": {
         "vendor": "google",
         "language": "de-DE",
         "label": "label1",
         "assemblyAiOptions": {
-          "apiKey": "apikey"
+          "apiKey": "apikey",
+          "serviceVersion": "v3",
+          "formatTurns": true,
+          "endOfTurnConfidenceThreshold": 0.5,
+          "minEndOfTurnSilenceWhenConfident": 500,
+          "maxTurnSilence": 2000
         }
       }
     },
@@ -207,6 +213,10 @@ test("validate correct verbs", async (t) => {
         "deepgramOptions": {
           "endpointing": 500,
           "noDelay": true,
+          "eotThreshold": 500,
+          "eotTimeoutMs": 5000,
+          "eagerEotThreshold": 200,
+          "mipOptOut": true
         }
       },
       "say": {
@@ -221,6 +231,10 @@ test("validate correct verbs", async (t) => {
           "fallbackVoice": "voice"
         }
       },
+      "say": {
+        "text": "To speak to Sales press 1 or say Sales.  To speak to customer support press 2 or say Support",
+        "instructions": "Voice: High-energy, upbeat, and encouraging, projecting enthusiasm and motivation."
+      },
       "say": {
         "text": "To speak to Sales press 1 or say Sales.  To speak to customer support press 2 or say Support",
         "synthesizer": {
@@ -265,6 +279,70 @@ test("validate correct verbs", async (t) => {
         }
       },
     },
+    {
+      "verb": "gather",
+      "actionHook": "http://example.com/collect",
+      "input": ["digits", "speech"],
+      "bargein": true,
+      "dtmfBargein": true,
+      "finishOnKey": "#",
+      "numDigits": 5,
+      "timeout": 8,
+      "recognizer": {
+        "vendor": "houndify",
+        "language": "en-US",
+        "houndifyOptions": {
+          "requestInfo": {
+            "Latitude": 30.6,
+            "Longitude": 30.6,
+            "City": "Huston"
+          },
+          "sampleRate": 16000,
+          "latitude": 30.6,
+          "longitude": 30.6,
+          "city": "Huston",
+          "state": "CA",
+          "country": "US",
+          "timeZone": "GMT",
+          "domain": "Voice",
+          "audioEndpoint": "https://api.houndify.com/v1/audio",
+          "maxSilenceSeconds": 5,
+          "maxSilenceAfterFullQuerySeconds": 3,
+          "maxSilenceAfterPartialQuerySeconds": 2,
+          "vadSensitivity": 0.5,
+          "vadTimeout": 1000,
+          "vadMode": "auto",
+          "vadVoiceMs": 250,
+          "vadSilenceMs": 500,
+          "vadDebug": true,
+          "audioFormat": "PCM16",
+          "enableNoiseReduction": true,
+          "enableProfanityFilter": false,
+          "enablePunctuation": true,
+          "enableCapitalization": true,
+          "confidenceThreshold": 0.7,
+          "enableDisfluencyFilter": true,
+          "maxResults": 5,
+          "enableWordTimestamps": true,
+          "maxAlternatives": 3,
+          "partialTranscriptInterval": 100,
+          "sessionTimeout": 30000,
+          "connectionTimeout": 5000,
+          "customVocabulary": ["jambonz", "telephony", "voip"],
+          "languageModel": "enhanced",
+          "audioQueryAbsoluteTimeout": 5
+        },
+        "gladiaOptions": {
+          "post_processing": {
+            "summarization": false,
+            "summarization_config": {
+              "type": "general"
+            },
+            "chapterization": false
+          },
+        }
+      },
+    },
     {
       "verb": "gather",
       "actionHook": "http://example.com/collect",
@@ -374,6 +452,10 @@ test("validate correct verbs", async (t) => {
        }
       }
     },
+    {
+      "verb": "config",
+      "autoStreamTts": true
+    },
     {
       "verb": "config",
       "vad": {
@@ -381,7 +463,28 @@ test("validate correct verbs", async (t) => {
         "voiceMs": 250,
         "silenceMs": 150,
         "strategy": "one-shot",
-        "mode": 2
+        "mode": 2,
+        "vendor": "webrtc",
+        "threshold": 0.5,
+        "speechPadMs": 1000
+      }
+    },
+    {
+      "verb": "config",
+      "noiseIsolation": {
+        "enable": true,
+        "vendor": "krisp",
+        "level": 3,
+        "model": "custom-model"
+      }
+    },
+    {
+      "verb": "config",
+      "turnTaking": {
+        "enable": true,
+        "vendor": "krisp",
+        "threshold": 0.5,
+        "model": "turn-taking-model"
       }
     },
     {
@@ -465,6 +568,24 @@ test("validate correct verbs", async (t) => {
         "interim": true
       }
     },
+    {
+      "verb": "transcribe",
+      "transcriptionHook": "http://example.com/transcribe",
+      "recognizer": {
+        "vendor": "nvidia",
+        "language" : "en-US",
+        "interim": true,
+        "elevenlabsOptions": {
+          "includeTimestamps": true,
+          "commitStrategy": "vad",
+          "vadSilenceThresholdSecs": 0.8,
+          "vadThreshold": 0.5,
+          "minSpeechDurationMs": 300,
+          "minSilenceDurationMs": 500,
+          "enableLogging": false
+        }
+      }
+    },
     {
       "verb": "transcribe",
       "transcriptionHook": "http://example.com/transcribe",
@@ -522,6 +643,94 @@ test("validate correct verbs", async (t) => {
       },
       "dual_streams": true,
       "timeLimit" : 10
+    },
+    {
+      "verb": "llm",
+      "vendor": 'ultravox',
+      "model": 'fixie-ai/ultravox',
+      "auth": {
+        "apiKey": "sk-1234567890abcdefg"
+      },
+      "llmOptions": {
+        "firstSpeaker": 'FIRST_SPEAKER_AGENT',
+      },
+      "mcpServers": [
+        {
+          "url": 'https://mcp.example.com',
+          "auth": {
+            "username": 'username',
+            "password": 'password'
+          }
+        }
+      ]
+    },
+    {
+      "verb": "s2s",
+      "vendor": "openai",
+      "llmOptions": {
+        "model": "gpt-4o-realtime"
+      }
+    },
+    {
+      "verb": "openai_s2s",
+      "llmOptions": {
+        "model": "gpt-4o-realtime"
+      }
+    },
+    {
+      "verb": "google_s2s",
+      "llmOptions": {
+        "model": "gemini-2.0-flash"
+      }
+    },
+    {
+      "verb": "elevenlabs_s2s",
+      "llmOptions": {
+        "agentId": "agent-123"
+      }
+    },
+    {
+      "verb": "stream",
+      "url": "wss://myrecorder.example.com/calls",
+      "mixType": "stereo"
+    },
+    {
+      "verb": "pipeline",
+      "stt": {
+        "vendor": "google",
+        "language": "en-US"
+      },
+      "tts": {
+        "vendor": "google",
+        "language": "en-US"
+      },
+      "llm": {
+        "vendor": "openai",
+        "llmOptions": {
+          "model": "gpt-4o"
+        }
+      },
+      "actionHook": "/pipeline/action",
+      "eventHook": "/pipeline/event",
+      "toolHook": "/pipeline/tool"
+    },
+    {
+      "verb": "transcribe",
+      "transcriptionHook": "http://example.com/transcribe",
+      "recognizer": {
+        "vendor": "speechmatics",
+        "language": "en",
+        "speechmaticsOptions": {
+          "transcription_config": {
+            "language": "en",
+            "enable_partials": true,
+            "max_delay": 2,
+            "conversation_config": {
+              "end_of_utterance_silence_trigger": 0.5
+            }
+          }
+        }
+      }
     }
   ];
   try {
@@ -551,4 +760,108 @@ test('invalid test', async (t) => {
   }
 
   t.end();
-})
+});
+
+test('verb synonyms: stream is synonym for listen', async (t) => {
+  // "verb" format
+  const result1 = normalizeJambones(logger, [
+    {"verb": "stream", "url": "wss://example.com/calls", "mixType": "stereo"}
+  ]);
+  t.equal(Object.keys(result1[0])[0], 'listen', 'stream verb is rewritten to listen');
+  t.equal(result1[0].listen.url, 'wss://example.com/calls', 'data is preserved');
+
+  // object-key format
+  const result2 = normalizeJambones(logger, [
+    {"stream": {"url": "wss://example.com/calls"}}
+  ]);
+  t.equal(Object.keys(result2[0])[0], 'listen', 'stream key is rewritten to listen');
+
+  // validate passes
+  try {
+    validate(logger, [{"verb": "stream", "url": "wss://example.com/calls"}]);
+    t.pass('stream verb validates successfully');
+  } catch (err) {
+    t.fail('stream verb should validate: ' + err);
+  }
+  t.end();
+});
+
+test('verb synonyms: s2s is synonym for llm', async (t) => {
+  const result = normalizeJambones(logger, [
+    {"verb": "s2s", "vendor": "openai", "llmOptions": {"model": "gpt-4o"}}
+  ]);
+  t.equal(Object.keys(result[0])[0], 'llm', 's2s verb is rewritten to llm');
+  t.equal(result[0].llm.vendor, 'openai', 'vendor is preserved');
+  t.equal(result[0].llm.llmOptions.model, 'gpt-4o', 'llmOptions preserved');
+
+  try {
+    validate(logger, [{"verb": "s2s", "vendor": "openai", "llmOptions": {"model": "gpt-4o"}}]);
+    t.pass('s2s verb validates successfully');
+  } catch (err) {
+    t.fail('s2s verb should validate: ' + err);
+  }
+  t.end();
+});
+
+test('vendor shortcuts: openai_s2s injects vendor', async (t) => {
+  const result = normalizeJambones(logger, [
+    {"verb": "openai_s2s", "llmOptions": {"model": "gpt-4o-realtime"}}
+  ]);
+  t.equal(Object.keys(result[0])[0], 'llm', 'openai_s2s is rewritten to llm');
+  t.equal(result[0].llm.vendor, 'openai', 'vendor is injected');
+  t.equal(result[0].llm.llmOptions.model, 'gpt-4o-realtime', 'llmOptions preserved');
+
+  try {
+    validate(logger, [{"verb": "openai_s2s", "llmOptions": {"model": "gpt-4o-realtime"}}]);
+    t.pass('openai_s2s validates successfully');
+  } catch (err) {
+    t.fail('openai_s2s should validate: ' + err);
+  }
+  t.end();
+});
+
+test('vendor shortcuts: all vendors work', async (t) => {
+  const vendors = [
+    'openai', 'microsoft', 'google', 'elevenlabs', 'deepgram', 'voiceagent', 'ultravox'
+  ];
+  for (const vendor of vendors) {
+    const verbName = `${vendor}_s2s`;
+    const result = normalizeJambones(logger, [
+      {"verb": verbName, "llmOptions": {}}
+    ]);
+    t.equal(Object.keys(result[0])[0], 'llm', `${verbName} rewrites to llm`);
+    t.equal(result[0].llm.vendor, vendor, `${verbName} injects vendor=${vendor}`);
+  }
+  t.end();
+});
+
+test('vendor shortcuts: object-key format works', async (t) => {
+  const result = normalizeJambones(logger, [
+    {"google_s2s": {"llmOptions": {"model": "gemini-2.0-flash"}}}
+  ]);
+  t.equal(Object.keys(result[0])[0], 'llm', 'google_s2s key is rewritten to llm');
+  t.equal(result[0].llm.vendor, 'google', 'vendor is injected');
+  t.equal(result[0].llm.llmOptions.model, 'gemini-2.0-flash', 'llmOptions preserved');
+  t.end();
+});
+
+test('vendor shortcuts: explicit vendor in data overrides injected vendor', async (t) => {
+  const result = normalizeJambones(logger, [
+    {"verb": "openai_s2s", "vendor": "custom", "llmOptions": {}}
+  ]);
+  t.equal(result[0].llm.vendor, 'custom', 'explicit vendor takes precedence');
+  t.end();
+});
+
+test('non-synonym verbs are unchanged', async (t) => {
+  const result = normalizeJambones(logger, [
+    {"verb": "say", "text": "hello"}
+  ]);
+  t.equal(Object.keys(result[0])[0], 'say', 'say verb is not transformed');
+
+  const result2 = normalizeJambones(logger, [
+    {"llm": {"vendor": "openai", "llmOptions": {}}}
+  ]);
+  t.equal(Object.keys(result2[0])[0], 'llm', 'llm verb is not transformed');
+  t.end();
+});