• Release Date: Nov 10, 2023
• Deployed in Helm chart release 0.1.33

• Add new speech vendors: Cobalt (STT), AssemblyAI (STT), Whisper (TTS), ElevenLabs (TTS)
• Support Docker containers for Azure speech
• Support custom voices on google
• Add ability to use a fallback speech vendor in case of error.
• Add support for speechSegmentationSilenceTimeoutMs for Microsoft Azure STT
• Optionally trim silence from Azure TTS
• Implement fastRecognitionTimeout to enable overriding speech vendor silence timeout used when determining the end of an utterance 448
• Add support for deepgram options smartFormatting and utteranceEndMs
• Calculate and show speech recognition latency in call recording view
• Add API to generate audio file from text using saved speech vendors 252
• Additional storage options for call recording: AWS compatible, Google cloud storage, Azure cloud storageq
• Add ability to pause and resume transcribe
• Add support for sending SIPREC to multiple SIPREC servers simultaneously
• Add continuous ASR support to transcribe verb.
• Allow an account to use multiple instances of the same speech vendor by assigning a unique label to each.
• Enable customized music on hold
• Allow use of tag in SIPREC application
• Allow custom headers to be passed when using SIPREC for recording
• Transcribe verb supports custom speech vendor
• Add option to prevent forwarding of P-Asserted-Identity header 460
• Add support for sending webhook requests through http proxy
• referHook on dial can now return an application which will be executed on the other leg (e.g. receive a REFER from A leg and run an application to dial the B leg to a new party C)
• Allow tag verb in waitHook response
• Allow pause verb in confirmHook on dial
• Integration with webrtc chrome extension to allow direct calling to apps, users, and queues
• Add option for padding crypto attributes in SDP when outdialing using SRTP
• Add support for speciyfing a redis sentinel password 239

• fix background listen request used for recording #391
• execute status callback async 394
• fix exception that appears in logs if session ends before last call status update
• fix snake case issue on customer data in webhooks 409
• outdial failure causes session to hang 411
• call status event was delayed when stopping background listen 413
• disable bi-directional audio on listen verbs used for call recording 415
• generate trace id before outdial so we can include it in custom header 418
• speech vendor missing in opentelemetry span 437
• fix bug with database caching 443
• caller should be able to enter more than one conference 350
• fixes from deepgram testing
• fix dialogflow TTS bug 518
• better logging of DNS errors
• Allow sip port to be null on outbound sip gateways, since this will trigger a DNS SRV request to determine ip and port
• Allow a top-level FQDN as an outbound sip gateway host 336

The following schema changes were made in this release:

DROP INDEX speech_credentials_idx_1 ON speech_credentials;
ALTER TABLE speech_credentials ADD COLUMN label VARCHAR(64);
ALTER TABLE applications ADD COLUMN speech_synthesis_label VARCHAR(64);
ALTER TABLE applications ADD COLUMN speech_recognizer_label VARCHAR(64);
ALTER TABLE applications ADD COLUMN use_for_fallback_speech BOOLEAN DEFAULT false;
ALTER TABLE applications ADD COLUMN fallback_speech_synthesis_vendor VARCHAR(64);
ALTER TABLE applications ADD COLUMN fallback_speech_synthesis_language VARCHAR(12);
ALTER TABLE applications ADD COLUMN fallback_speech_synthesis_voice VARCHAR(64);
ALTER TABLE applications ADD COLUMN fallback_speech_synthesis_label VARCHAR(64);
ALTER TABLE applications ADD COLUMN fallback_speech_recognizer_vendor VARCHAR(64);
ALTER TABLE applications ADD COLUMN fallback_speech_recognizer_language VARCHAR(64);
ALTER TABLE applications ADD COLUMN fallback_speech_recognizer_label VARCHAR(64);
ALTER TABLE sip_gateways ADD COLUMN pad_crypto BOOLEAN NOT NULL DEFAULT 0;
ALTER TABLE sip_gateways MODIFY port INTEGER;
CREATE TABLE google_custom_voices
(
google_custom_voice_sid CHAR(36) NOT NULL UNIQUE ,
speech_credential_sid CHAR(36) NOT NULL,
model VARCHAR(512) NOT NULL,
reported_usage ENUM('REPORTED_USAGE_UNSPECIFIED','REALTIME','OFFLINE') DEFAULT 'REALTIME',
name VARCHAR(64) NOT NULL,
PRIMARY KEY (google_custom_voice_sid)
);
CREATE INDEX google_custom_voice_sid_idx ON google_custom_voices (google_custom_voice_sid);
CREATE INDEX speech_credential_sid_idx ON google_custom_voices (speech_credential_sid);
ALTER TABLE google_custom_voices ADD FOREIGN KEY speech_credential_sid_idxfk (speech_credential_sid) REFERENCES speech_credentials (speech_credential_sid) ON DELETE CASCADE;
ALTER TABLE clients ADD COLUMN allow_direct_queue_calling BOOLEAN NOT NULL DEFAULT 1;
ALTER TABLE clients ADD COLUMN allow_direct_user_calling BOOLEAN NOT NULL DEFAULT 1;
ALTER TABLE clients ADD COLUMN allow_direct_app_calling BOOLEAN NOT NULL DEFAULT 1;

• Available shortly on AWS Marketplace
• Deploy to Kubernetes using this Helm chart

Questions? Contact us at support@jambonz.org