0.0.48

Feat/mod deepgram tts

lib/synth-audio.js

@@ -141,12 +141,12 @@ async function synthAudio(client, logger, stats, { account_sid,
   let filePath;
   if (['nuance', 'nvidia'].includes(vendor) ||
     (
-      process.env.JAMBONES_TTS_TRIM_SILENCE &&
+      (process.env.JAMBONES_TTS_TRIM_SILENCE || !process.env.JAMBONES_DISABLE_TTS_STREAMING) &&
       ['microsoft', 'azure'].includes(vendor)
     ) ||
     (
       !process.env.JAMBONES_DISABLE_TTS_STREAMING &&
-      vendor === 'elevenlabs'
+      ['elevenlabs', 'deepgram'].includes(vendor)
     )
   ) {
     filePath = `${TMP_FOLDER}/${key.replace('tts:', `tts-${salt || ''}`)}.r8`;
@@ -212,7 +212,9 @@ async function synthAudio(client, logger, stats, { account_sid,
         if (audioBuffer?.filePath) return audioBuffer;
         break;
       case 'deepgram':
-        audioBuffer = await synthDeepgram(logger, {credentials, stats, model, text});
+        audioBuffer = await synthDeepgram(logger, {credentials, stats, model, text,
+          renderForCaching, disableTtsStreaming});
+        if (audioBuffer?.filePath) return audioBuffer;
         break;
       case vendor.startsWith('custom') ? vendor : 'cant_match_value':
         ({ audioBuffer, filePath } = await synthCustomVendor(logger,
@@ -412,6 +414,7 @@ const synthMicrosoft = async(logger, {
       params += `,language=${language}`;
       params += ',vendor=microsoft';
       params += `,voice=${voice}`;
+      params += ',write_cache_file=1';
       if (region) params += `,region=${region}`;
       if (custom_tts_endpoint) params += `,endpointId=${custom_tts_endpoint}`;
       if (process.env.JAMBONES_HTTP_PROXY_IP) params += `,http_proxy_ip=${process.env.JAMBONES_HTTP_PROXY_IP}`;
@@ -719,8 +722,22 @@ const synthWhisper = async(logger, {credentials, stats, voice, text, renderForCa
   }
 };
 
-const synthDeepgram = async(logger, {credentials, stats, model, text}) => {
+const synthDeepgram = async(logger, {credentials, stats, model, text, renderForCaching, disableTtsStreaming}) => {
   const {api_key} = credentials;
+  if (!process.env.JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+    let params = '';
+    params += `{api_key=${api_key}`;
+    params += ',vendor=deepgram';
+    params += `,voice=${model}`;
+    params += ',write_cache_file=1';
+    params += '}';
+
+    return {
+      filePath: `say:${params}${text.replace(/\n/g, ' ')}`,
+      servedFromCache: false,
+      rtt: 0
+    };
+  }
   try {
     const post = bent('https://api.deepgram.com', 'POST', 'buffer', {
       'Authorization': `Token ${api_key}`,

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.0.45",
+  "version": "0.0.48",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "@jambonz/speech-utils",
-      "version": "0.0.45",
+      "version": "0.0.48",
       "license": "MIT",
       "dependencies": {
         "@aws-sdk/client-polly": "^3.496.0",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.0.45",
+  "version": "0.0.48",
   "description": "TTS-related speech utilities for jambonz",
   "main": "index.js",
   "author": "Dave Horton",

test/synth.js

@@ -595,6 +595,7 @@ test('Deepgram speech synth tests', async(t) => {
       },
       model: 'aura-asteria-en',
       text,
+      renderForCaching: true
     });
     t.ok(!opts.servedFromCache, `successfully synthesized deepgram audio to ${opts.filePath}`);