add support for speechmatics languages and voices (#355)

* support stt speechmatics

* support speechmatics region authentication

* update testcase for speechmatics_stt_uri

db/jambones-sql.sql

@@ -162,7 +162,7 @@ regex VARCHAR(32) NOT NULL COMMENT 'regex-based pattern match against dialed num
 description VARCHAR(1024),
 priority INTEGER NOT NULL COMMENT 'lower priority routes are attempted first',
 PRIMARY KEY (lcr_route_sid)
-) COMMENT='An ordered list of  digit patterns in an LCR table.  The pat';
+) COMMENT='An ordered list of  digit patterns in an LCR table.  The patterns are tested in sequence until one matches';
 
 CREATE TABLE lcr
 (
@@ -173,7 +173,7 @@ default_carrier_set_entry_sid CHAR(36) COMMENT 'default carrier/route to use whe
 service_provider_sid CHAR(36),
 account_sid CHAR(36),
 PRIMARY KEY (lcr_sid)
-) COMMENT='An LCR (least cost routing) table that is used by a service ';
+) COMMENT='An LCR (least cost routing) table that is used by a service provider or account to make decisions about routing outbound calls when multiple carriers are available.';
 
 CREATE TABLE password_settings
 (
@@ -359,7 +359,8 @@ CREATE TABLE system_information
 domain_name VARCHAR(255),
 sip_domain_name VARCHAR(255),
 monitoring_domain_name VARCHAR(255),
-private_network_cidr VARCHAR(8192)
+private_network_cidr VARCHAR(8192),
+log_level ENUM('info', 'debug') NOT NULL DEFAULT 'info'
 );
 
 CREATE TABLE users
@@ -553,6 +554,7 @@ siprec_hook_sid CHAR(36),
 record_all_calls BOOLEAN NOT NULL DEFAULT false,
 record_format VARCHAR(16) NOT NULL DEFAULT 'mp3',
 bucket_credential VARCHAR(8192) COMMENT 'credential used to authenticate with storage service',
+enable_debug_log BOOLEAN NOT NULL DEFAULT false,
 PRIMARY KEY (account_sid)
 ) COMMENT='An enterprise that uses the platform for comm services';
 

db/upgrade-jambonz-db.js

@@ -200,6 +200,8 @@ const sql = {
   ],
   9002: [
     'ALTER TABLE system_information ADD COLUMN private_network_cidr VARCHAR(8192)',
+    'ALTER TABLE system_information ADD COLUMN log_level ENUM(\'info\', \'debug\') NOT NULL DEFAULT \'info\'',
+    'ALTER TABLE accounts ADD COLUMN enable_debug_log BOOLEAN NOT NULL DEFAULT false',
   ]
 };
 

lib/routes/api/speech-credentials.js

@@ -10,7 +10,8 @@ const {decryptCredential, testWhisper, testDeepgramTTS,
   testPlayHT,
   testRimelabs,
   testVerbioTts,
-  testVerbioStt} = require('../../utils/speech-utils');
+  testVerbioStt,
+  testSpeechmaticsStt} = require('../../utils/speech-utils');
 const {DbErrorUnprocessableRequest, DbErrorForbidden, DbErrorBadRequest} = require('../../utils/errors');
 const {
   testGoogleTts,
@@ -122,6 +123,7 @@ const encryptCredential = (obj) => {
     secret,
     nuance_tts_uri,
     nuance_stt_uri,
+    speechmatics_stt_uri,
     deepgram_stt_uri,
     deepgram_stt_use_tls,
     deepgram_tts_uri,
@@ -236,6 +238,12 @@ const encryptCredential = (obj) => {
       const elevenlabsData = JSON.stringify({api_key, model_id, options});
       return encrypt(elevenlabsData);
 
+    case 'speechmatics':
+      assert(api_key, 'invalid speechmatics speech credential: api_key is required');
+      assert(speechmatics_stt_uri, 'invalid speechmatics speech credential: speechmatics_stt_uri is required');
+      const speechmaticsData = JSON.stringify({api_key, speechmatics_stt_uri, options});
+      return encrypt(speechmaticsData);
+
     case 'playht':
       assert(api_key, 'invalid playht speech credential: api_key is required');
       assert(user_id, 'invalid playht speech credential: user_id is required');
@@ -768,6 +776,18 @@ router.get('/:sid/test', async(req, res) => {
           SpeechCredential.ttsTestResult(sid, false);
         }
       }
+    } else if (cred.vendor === 'speechmatics') {
+      const {api_key} = credential;
+      if (cred.use_for_stt) {
+        try {
+          await testSpeechmaticsStt(logger, {api_key});
+          results.stt.status = 'ok';
+          SpeechCredential.ttsTestResult(sid, true);
+        } catch (err) {
+          results.stt = {status: 'fail', reason: err.message};
+          SpeechCredential.ttsTestResult(sid, false);
+        }
+      }
     } else if (cred.vendor === 'playht') {
       if (cred.use_for_tts) {
         try {

lib/utils/speech-data/stt-speechmatics.js

@@ -0,0 +1,218 @@
+module.exports = [
+  {
+    name: 'Automatic',
+    value: 'auto',
+  },
+  {
+    name: 'Arabic',
+    value: 'ar',
+  },
+  {
+    name: 'Bashkir',
+    value: 'ba',
+  },
+  {
+    name: 'Basque',
+    value: 'eu',
+  },
+  {
+    name: 'Belarusian',
+    value: 'be',
+  },
+  {
+    name: 'Bulgarian',
+    value: 'bg',
+  },
+  {
+    name: 'Cantonese',
+    value: 'yue',
+  },
+  {
+    name: 'Catalan',
+    value: 'ca',
+  },
+  {
+    name: 'Croatian',
+    value: 'hr',
+  },
+  {
+    name: 'Czech',
+    value: 'cs',
+  },
+  {
+    name: 'Danish',
+    value: 'da',
+  },
+  {
+    name: 'Dutch',
+    value: 'nl',
+  },
+  {
+    name: 'English',
+    value: 'en',
+  },
+  {
+    name: 'Esperanto',
+    value: 'eo',
+  },
+  {
+    name: 'Estonian',
+    value: 'et',
+  },
+  {
+    name: 'Finnish',
+    value: 'fi',
+  },
+  {
+    name: 'French',
+    value: 'fr',
+  },
+  {
+    name: 'Galician',
+    value: 'gl',
+  },
+  {
+    name: 'German',
+    value: 'de',
+  },
+  {
+    name: 'Greek',
+    value: 'el',
+  },
+  {
+    name: 'Hebrew',
+    value: 'he',
+  },
+  {
+    name: 'Hindi',
+    value: 'hi',
+  },
+  {
+    name: 'Hungarian',
+    value: 'hu',
+  },
+  {
+    name: 'Irish',
+    value: 'ga',
+  },
+  {
+    name: 'Interlingua',
+    value: 'ia',
+  },
+  {
+    name: 'Italian',
+    value: 'it',
+  },
+  {
+    name: 'Indonesian',
+    value: 'id',
+  },
+  {
+    name: 'Japanese',
+    value: 'ja',
+  },
+  {
+    name: 'Korean',
+    value: 'ko',
+  },
+  {
+    name: 'Latvian',
+    value: 'lv',
+  },
+  {
+    name: 'Lithuanian',
+    value: 'lt',
+  },
+  {
+    name: 'Maltese',
+    value: 'mt',
+  },
+  {
+    name: 'Malay',
+    value: 'ms',
+  },
+  {
+    name: 'Mandarin',
+    value: 'cmn',
+  },
+  {
+    name: 'Marathi',
+    value: 'mr',
+  },
+  {
+    name: 'Mongolian',
+    value: 'mn',
+  },
+  {
+    name: 'Norwegian',
+    value: 'no',
+  },
+  {
+    name: 'Persian',
+    value: 'fa',
+  },
+  {
+    name: 'Polish',
+    value: 'pl',
+  },
+  {
+    name: 'Portuguese',
+    value: 'pt',
+  },
+  {
+    name: 'Romanian',
+    value: 'ro',
+  },
+  {
+    name: 'Russian',
+    value: 'ru',
+  },
+  {
+    name: 'Slovakian',
+    value: 'sk',
+  },
+  {
+    name: 'Slovenian',
+    value: 'sl',
+  },
+  {
+    name: 'Spanish',
+    value: 'es',
+  },
+  {
+    name: 'Spanish & English bilingual',
+    value: 'es',
+  },
+  {
+    name: 'Swedish',
+    value: 'sv',
+  },
+  {
+    name: 'Tamil',
+    value: 'ta',
+  },
+  {
+    name: 'Thai',
+    value: 'th',
+  },
+  {
+    name: 'Turkish',
+    value: 'tr',
+  },
+  {
+    name: 'Uyghur',
+    value: 'ug',
+  },
+  {
+    name: 'Ukrainian',
+    value: 'uk',
+  },
+  {
+    name: 'Vietnamese',
+    value: 'vi',
+  },
+  {
+    name: 'Welsh',
+    value: 'cy',
+  },
+];

lib/utils/speech-data/tts-model-playht.js

@@ -1,4 +1,5 @@
 module.exports = [
+  { name: 'Play3.0', value: 'Play3.0' },
   { name: 'PlayHT2.0-turbo', value: 'PlayHT2.0-turbo' },
   { name: 'PlayHT2.0', value: 'PlayHT2.0' },
   { name: 'PlayHT1.0', value: 'PlayHT1.0' },

lib/utils/speech-utils.js

@@ -7,6 +7,7 @@ const bent = require('bent');
 const fs = require('fs');
 const { AssemblyAI } = require('assemblyai');
 const {decrypt, obscureKey} = require('./encrypt-decrypt');
+const { RealtimeSession } = require('speechmatics');
 
 const TtsGoogleLanguagesVoices = require('./speech-data/tts-google');
 const TtsAwsLanguagesVoices = require('./speech-data/tts-aws');
@@ -35,6 +36,7 @@ const SttIbmLanguagesVoices = require('./speech-data/stt-ibm');
 const SttNvidiaLanguagesVoices = require('./speech-data/stt-nvidia');
 const SttCobaltLanguagesVoices = require('./speech-data/stt-cobalt');
 const SttSonioxLanguagesVoices = require('./speech-data/stt-soniox');
+const SttSpeechmaticsLanguagesVoices = require('./speech-data/stt-speechmatics');
 const SttAssemblyaiLanguagesVoices = require('./speech-data/stt-assemblyai');
 const SttVerbioLanguagesVoices = require('./speech-data/stt-verbio');
 
@@ -54,6 +56,61 @@ const testSonioxStt = async(logger, credentials) => {
   });
 };
 
+const testSpeechmaticsStt = async(logger, credentials) => {
+  const {api_key, speechmatics_stt_uri} = credentials;
+  return new Promise(async(resolve, reject) => {
+    try {
+      const session = new RealtimeSession({ apiKey: api_key, realtimeUrl: speechmatics_stt_uri });
+      let transcription = '';
+      session.addListener('Error', (error) => {
+        reject(error);
+      });
+
+      session.addListener('AddTranscript', (message) => {
+        transcription += message.metadata.transcript;
+      });
+
+      session.addListener('EndOfTranscript', () => {
+        resolve(transcription);
+      });
+
+      session
+        .start({
+          transcription_config: {
+            language: 'en',
+            operating_point: 'enhanced',
+            enable_partials: true,
+            max_delay: 2,
+          },
+          audio_format: { type: 'file' },
+        })
+        .then(() => {
+          //prepare file stream
+          const fileStream = fs.createReadStream(`${__dirname}/../../data/test_audio.wav`);
+
+          //send it
+          fileStream.on('data', (sample) => {
+            session.sendAudio(sample);
+          });
+
+          //end the session
+          fileStream.on('end', () => {
+            session.stop();
+          });
+
+          return;
+
+        })
+        .catch((error) => {
+          reject(error);
+        });
+    } catch (error) {
+      logger.info({error}, 'failed to get speechmatics transcript');
+      reject(error);
+    }
+  });
+};
+
 const testNuanceTts = async(logger, getTtsVoices, credentials) => {
   const voices = await getTtsVoices({vendor: 'nuance', credentials});
   return voices;
@@ -532,6 +589,10 @@ function decryptCredential(obj, credential, logger, isObscureKey = true) {
   } else if ('soniox' === obj.vendor) {
     const o = JSON.parse(decrypt(credential));
     obj.api_key = isObscureKey ? obscureKey(o.api_key) : o.api_key;
+  } else if ('speechmatics' === obj.vendor) {
+    const o = JSON.parse(decrypt(credential));
+    obj.api_key = isObscureKey ? obscureKey(o.api_key) : o.api_key;
+    obj.speechmatics_stt_uri = o.speechmatics_stt_uri;
   } else if ('elevenlabs' === obj.vendor) {
     const o = JSON.parse(decrypt(credential));
     obj.api_key = isObscureKey ? obscureKey(o.api_key) : o.api_key;
@@ -623,6 +684,8 @@ async function getLanguagesAndVoicesForVendor(logger, vendor, credential, getTts
       return await getLanguagesVoicesForWhisper(credential, getTtsVoices, logger);
     case 'verbio':
       return await getLanguagesVoicesForVerbio(credential, getTtsVoices, logger);
+    case 'speechmatics':
+      return await getLanguagesVoicesForSpeechmatics(credential, getTtsVoices, logger);
     default:
       logger.info(`invalid vendor ${vendor}, return empty result`);
       throw new Error(`Invalid vendor ${vendor}`);
@@ -735,6 +798,10 @@ async function getLanguagesVoicesForSoniox(credential) {
   return tranform(undefined, SttSonioxLanguagesVoices);
 }
 
+async function getLanguagesVoicesForSpeechmatics(credential) {
+  return tranform(undefined, SttSpeechmaticsLanguagesVoices);
+}
+
 async function getLanguagesVoicesForElevenlabs(credential) {
   if (credential) {
     const get = bent('https://api.elevenlabs.io', 'GET', 'json', {
@@ -816,7 +883,10 @@ async function getLanguagesVoicesForPlayHT(credential) {
     };
 
     const ttsVoices = list_voices.reduce((acc, voice) => {
-      if (!credential.voice_engine.includes(voice.voice_engine)) {
+      // Play3.0 support all voice for PlayHT2.0*
+      const filteredVoiceEngine = credential.voice_engine === 'Play3.0' ?
+        `${credential.voice_engine}_PlayHT2.0_PlayHT2.0-turbo` : credential.voice_engine;
+      if (!filteredVoiceEngine.includes(voice.voice_engine)) {
         return acc;
       }
       const languageCode = voice.language_code;
@@ -1063,5 +1133,6 @@ module.exports = {
   testWhisper,
   testVerbioTts,
   testVerbioStt,
-  getLanguagesAndVoicesForVendor
+  getLanguagesAndVoicesForVendor,
+  testSpeechmaticsStt
 };

package.json

@@ -28,8 +28,8 @@
     "@jambonz/db-helpers": "^0.9.3",
     "@jambonz/lamejs": "^1.2.2",
     "@jambonz/mw-registrar": "^0.2.7",
-    "@jambonz/realtimedb-helpers": "^0.8.9",
-    "@jambonz/speech-utils": "^0.1.13",
+    "@jambonz/realtimedb-helpers": "^0.8.10",
+    "@jambonz/speech-utils": "^0.1.18",
     "@jambonz/time-series": "^0.2.8",
     "@jambonz/verb-specifications": "^0.0.72",
     "@soniox/soniox-node": "^1.2.2",
@@ -52,6 +52,7 @@
     "passport-http-bearer": "^1.0.1",
     "pino": "^8.20.0",
     "short-uuid": "^4.2.2",
+    "speechmatics": "^4.0.0",
     "stream-buffers": "^3.0.2",
     "stripe": "^14.24.0",
     "swagger-ui-express": "^5.0.0",

test/accounts.js

@@ -152,7 +152,9 @@ test('account tests', async(t) => {
       auth: authAdmin,
       json: true,
     });
+    console.log(result);
     t.ok(result.name === 'daveh' , 'successfully retrieved account by sid');
+    t.ok(result.enable_debug_log  === 0 , 'enable_debug_log default value ok');
 
     /* update account with account level token */
     result = await request.put(`/Accounts/${sid}`, {
@@ -177,8 +179,8 @@ test('account tests', async(t) => {
           name: 'recordings',
           access_key_id: 'access_key_id',
           secret_access_key: 'secret access key'
-        }
-        
+        },
+        enable_debug_log: true
       }
     });
     t.ok(result.statusCode === 204, 'successfully updated account using account level token');
@@ -194,6 +196,7 @@ test('account tests', async(t) => {
     t.ok(result.bucket_credential.access_key_id === 'access_key_id', 'bucket_access_key_id was updated');
     t.ok(result.record_all_calls === 1, 'record_all_calls was updated');
     t.ok(result.record_format === 'wav', 'record_format was updated');
+    t.ok(result.enable_debug_log, 'enable_debug_log was updated');
 
     /* verify that account level api key last_used was updated*/
     result = await request.get(`/Accounts/${sid}/ApiKeys`, {

test/speech-credentials.js

@@ -532,6 +532,39 @@ test('speech credentials tests', async(t) => {
       t.ok(result.statusCode === 204, 'successfully deleted speech credential');
     }
 
+    /* add a credential for Speechmatics */
+    if (process.env.SPEECHMATICS_API_KEY) {
+      result = await request.post(`/Accounts/${account_sid}/SpeechCredentials`, {
+        resolveWithFullResponse: true,
+        auth: authUser,
+        json: true,
+        body: {
+          vendor: 'speechmatics',
+          use_for_stt: true,
+          api_key: process.env.SPEECHMATICS_API_KEY,
+          speechmatics_stt_uri: 'eu2.rt.speechmatics.com'
+        }
+      });
+      t.ok(result.statusCode === 201, 'successfully added speech credential for speechmatics');
+      const ms_sid = result.body.sid;
+
+      /* test the speech credential */
+      result = await request.get(`/Accounts/${account_sid}/SpeechCredentials/${ms_sid}/test`, {
+        resolveWithFullResponse: true,
+        auth: authUser,
+        json: true,   
+      });
+      console.log(JSON.stringify(result));
+      t.ok(result.statusCode === 200 && result.body.stt.status === 'ok', 'successfully tested speech credential for speechmatics');
+
+      /* delete the credential */
+      result = await request.delete(`/Accounts/${account_sid}/SpeechCredentials/${ms_sid}`, {
+        auth: authUser,
+        resolveWithFullResponse: true,
+      });
+      t.ok(result.statusCode === 204, 'successfully deleted speech credential');
+    }
+
     /* add a credential for nvidia */
     result = await request.post(`/Accounts/${account_sid}/SpeechCredentials`, {
       resolveWithFullResponse: true,

test/system-information.js

@@ -17,7 +17,8 @@ test('system information test', async(t) => {
         domain_name: 'test.com',
         sip_domain_name: 'sip.test.com',
         monitoring_domain_name: 'monitor.test.com',
-        private_network_cidr: '192.168.1.0/24, 10.10.100.1'
+        private_network_cidr: '192.168.1.0/24, 10.10.100.1',
+        log_level: 'info'
       }
     });
     t.ok(result.statusCode === 201, 'successfully created system information ');
@@ -26,6 +27,7 @@ test('system information test', async(t) => {
     t.ok(body.sip_domain_name === 'sip.test.com', 'added sip_domain_name ok');
     t.ok(body.monitoring_domain_name === 'monitor.test.com', 'added monitoring_domain_name ok');
     t.ok(body.private_network_cidr === '192.168.1.0/24, 10.10.100.1', 'added private_network_cidr ok');
+    t.ok(body.log_level === 'info', 'added log_level ok');
 
     result = await request.get('/SystemInformation', {
       auth: authAdmin,
@@ -35,6 +37,7 @@ test('system information test', async(t) => {
     t.ok(result.sip_domain_name === 'sip.test.com', 'get sip_domain_name ok');
     t.ok(result.monitoring_domain_name === 'monitor.test.com', 'get monitoring_domain_name ok');
     t.ok(result.private_network_cidr === '192.168.1.0/24, 10.10.100.1', 'get private_network_cidr ok'); 
+    t.ok(result.log_level === 'info', 'added log_level ok');
 
     result = await request.post('/SystemInformation', {
       resolveWithFullResponse: true,
@@ -44,7 +47,8 @@ test('system information test', async(t) => {
         domain_name: 'test1.com',
         sip_domain_name: 'sip1.test.com',
         monitoring_domain_name: 'monitor1.test.com',
-        private_network_cidr: ''
+        private_network_cidr: '',
+        log_level: 'debug'
       }
     });
     t.ok(result.statusCode === 201, 'successfully updated system information ');
@@ -53,6 +57,7 @@ test('system information test', async(t) => {
     t.ok(body.sip_domain_name === 'sip1.test.com', 'updated sip_domain_name ok');
     t.ok(body.monitoring_domain_name === 'monitor1.test.com', 'updated monitoring_domain_name ok');
     t.ok(body.private_network_cidr === '', 'updated private_network_cidr ok');
+    t.ok(body.log_level === 'debug', 'updated log_level ok');
 
     result = await request.get('/SystemInformation', {
       auth: authAdmin,
@@ -61,6 +66,7 @@ test('system information test', async(t) => {
     t.ok(result.domain_name === 'test1.com', 'get domain_name ok');
     t.ok(result.sip_domain_name === 'sip1.test.com', 'get sip_domain_name ok');
     t.ok(result.monitoring_domain_name === 'monitor1.test.com', 'get monitoring_domain_name ok');
+    t.ok(result.log_level === 'debug', 'updated log_level ok');
 
   } catch(err) {
     console.error(err);