diff --git a/package.json b/package.json
index 00b02e00..cbd7f862 100644
--- a/package.json
+++ b/package.json
@@ -85,7 +85,6 @@
     "ibm-watson": "6.1.1",
     "js-beautify": "1.13.13",
     "marked": "2.0.7",
-    "microsoft-cognitiveservices-speech-sdk": "1.14.1",
     "ms-rest-azure": "3.0.0",
     "nexmo": "2.9.1",
     "node-cron": "3.0.0",
diff --git a/packages/core.gbapp/services/GBConversationalService.ts b/packages/core.gbapp/services/GBConversationalService.ts
index 8bc49fba..f833470b 100644
--- a/packages/core.gbapp/services/GBConversationalService.ts
+++ b/packages/core.gbapp/services/GBConversationalService.ts
@@ -297,53 +297,6 @@ export class GBConversationalService {
     await min.whatsAppDirectLine.sendToDevice(mobile, message);
   }
 
-  public static async getAudioBufferFromText(speechKey, cloudRegion, text, locale): Promise<string> {
-    return new Promise<string>(async (resolve, reject) => {
-      const name = GBAdminService.getRndReadableIdentifier();
-
-      const waveFilename = `work/tmp${name}.pcm`;
-      const sdk = require('microsoft-cognitiveservices-speech-sdk');
-      sdk.Recognizer.enableTelemetry(false);
-
-      var audioConfig = sdk.AudioConfig.fromAudioFileOutput(waveFilename);
-      var speechConfig = sdk.SpeechConfig.fromSubscription(speechKey, cloudRegion);
-
-      var synthesizer = new sdk.SpeechSynthesizer(speechConfig, audioConfig);
-
-      try {
-        speechConfig.speechSynthesisLanguage = locale;
-        speechConfig.speechSynthesisVoiceName = 'pt-BR-FranciscaNeural';
-
-        synthesizer.speakTextAsync(text, result => {
-          if (result.reason === sdk.ResultReason.SynthesizingAudioCompleted) {
-            let raw = Buffer.from(result.audioData);
-            fs.writeFileSync(waveFilename, raw);
-            GBLog.info(`Audio data byte size: ${result.audioData.byteLength}.`);
-            const oggFilenameOnly = `tmp${name}.ogg`;
-            const oggFilename = `work/${oggFilenameOnly}`;
-
-            const output = fs.createWriteStream(oggFilename);
-            const transcoder = new prism.FFmpeg({
-              args: ['-analyzeduration', '0', '-loglevel', '0', '-f', 'opus', '-ar', '16000', '-ac', '1']
-            });
-
-            fs.createReadStream(waveFilename).pipe(transcoder).pipe(output);
-
-            let url = urlJoin(GBServer.globals.publicAddress, 'audios', oggFilenameOnly);
-            resolve(url);
-          } else {
-            const error = 'Speech synthesis canceled, ' + result.errorDetails;
-            reject(error);
-          }
-          synthesizer.close();
-          synthesizer = undefined;
-        });
-      } catch (error) {
-        reject(error);
-      }
-    });
-  }
-
   public static async getTextFromAudioBuffer(speechKey, cloudRegion, buffer, locale): Promise<string> {
     return new Promise<string>(async (resolve, reject) => {
       try {
diff --git a/packages/whatsapp.gblib/services/WhatsappDirectLine.ts b/packages/whatsapp.gblib/services/WhatsappDirectLine.ts
index d7fe8dcd..01c3e782 100644
--- a/packages/whatsapp.gblib/services/WhatsappDirectLine.ts
+++ b/packages/whatsapp.gblib/services/WhatsappDirectLine.ts
@@ -443,12 +443,13 @@ export class WhatsappDirectLine extends GBService {
 
   public async sendTextAsAudioToDevice(to, msg) {
 
-    const url = await GBConversationalService.getAudioBufferFromText(
-      this.min.instance.speechKey,
-      this.min.instance.cloudLocation,
-      msg, this.locale
-    );
-
+    // DISABLED:
+    // const url = await GBConversationalService.getAudioBufferFromText(
+    //   this.min.instance.speechKey,
+    //   this.min.instance.cloudLocation,
+    //   msg, this.locale
+    // );
+    const url = 'https://';
     await this.sendFileToDevice(to, url, 'Audio', msg);
   }