This guide explains production-tested fixes for ASR and TTS issues on the Kore.ai Voice Gateway. Each section identifies the symptom, root cause, and the exact configuration or code change required.

All parameters are applied using:

userSessionUtils.setCallControlParam('paramName', value);

Microsoft Azure ASR

Utterance Splitting — Speech Split into Multiple Parts

Symptom

The ASR splits one continuous user utterance into multiple turns. For example, "Hello, I want to check my balance" arrives as two separate inputs: "Hello" followed by "I want to check my balance". This causes intent mismatches, loop behaviour, or the bot responding to partial input.

Root Cause

Azure Speech uses AzureSegmentationSilenceTimeoutMs to detect end-of-speech. Natural pauses of 300 ms or more between words can trigger a premature commit. The default value is too aggressive for conversational IVR use cases.

Fix

Add a Script node at the start of your Experience Flow:

// Standard conversational flows
userSessionUtils.setCallControlParam(
  'AzureSegmentationSilenceTimeoutMs', 1200
);

// Longer conversational utterances (agentic app flows)
userSessionUtils.setCallControlParam(
  'AzureSegmentationSilenceTimeoutMs', 2000
);

// Do NOT exceed 3000 — significantly increases latency.

Parameter Reference

Parameter	Value / Range	Purpose
AzureSegmentationSilenceTimeoutMs	800–2000 ms	End-of-speech silence window. Lower = faster response. Higher = fewer splits.
sttMinConfidence	0.4–0.6	Lower to accept more inputs; raise to filter background noise.
botNoInputTimeoutMS	5000–8000 ms	Increase for non-English speakers or callers who pause before speaking.

Digit and Number Misrecognition

Symptom

Numbers are misinterpreted by the ASR engine. Common patterns:

• Spoken Hindi numbers like ek (one) combined with 22,000 become 1,202,000
• Card numbers with repeated digits are truncated or garbled
• Confidence scores are above threshold (for example, 75.7%), but the transcription is still incorrect

Fix

// Enable continuous ASR at the digit-collection entity node
userSessionUtils.setCallControlParam('continuousASR', true);
userSessionUtils.setCallControlParam('continuousASRTimeoutInMS', 2000);

// DTMF fallback — strongly recommended for 16-digit card numbers
userSessionUtils.setCallControlParam('dtmfCollectMaxDigits', 16);
userSessionUtils.setCallControlParam('dtmfCollectInterDigitTimeoutMS', 4000);

Input Type Guidance

Input Type	Recommended Method	Configuration Notes	Error Rate
4–8 digit PIN	Speech ASR	Enable continuousASR.	Low
10-digit phone number	Speech or DTMF	Speech for English; DTMF for non-English.	Medium
16-digit card number	DTMF preferred	ASR struggles with repeated digits. Use DTMF as primary.	High with ASR
Currency amounts (Hindi)	Speech + lower threshold	sttMinConfidence=0.4 + AzureSegmentationSilenceTimeoutMs=1500	Medium

---

Widespread ASR Failure — Outage or Concurrency Breach

Symptom

All IVR calls fail simultaneously. The bot responds with the default error message, and no speech is recognised across any bots. An Azure region outage typically causes either a concurrency limit breach or expired API credentials.

Immediate Recovery Steps

1. Check Azure Service Health — Verify the Azure region status at the Azure Service Health dashboard. Confirm whether the outage is regional or global.
2. Verify API Credentials — Confirm the Azure Speech API key and endpoint have not expired. Rotate credentials if necessary and update them in the Kore.ai platform.
3. Switch to Fallback ASR — Activate the fallback ASR configuration (different region or secondary vendor) to restore service while the primary region recovers.
4. Monitor Concurrency Limits — Review your Azure subscription concurrency quota. If the limit has been breached, request a quota increase or shed load to the fallback.

Prevention — Configure Before Go-Live

// Primary and Fallback MUST use different labels.
// Using the same label for both provides no fallback protection.

Primary ASR:  Label = your-primary-label    // primary region
Fallback ASR: Label = your-fallback-label   // different region, same vendor

Microsoft Azure TTS

Mispronounced Words — SSML Phoneme and Say-As Tags

Symptom

Azure Neural Voice mispronounces product names, currency amounts, acronyms, or non-English terms. Digits may be read as words — for example, 6987 spoken as "Che Nau Sath Aath" when using a mismatched voice.

Fix

Use SSML tags in your Message nodes:

<!-- Currency amounts -->
<speak>
  <say-as interpret-as="currency" language="hi-IN">
    ₹3,40,650.23
  </say-as>
</speak>

<!-- Digit-by-digit — prevents misread artifacts -->
<speak>
  Your OTP is <say-as interpret-as="digits">6987</say-as>
</speak>

<!-- Acronym spelled out -->
<speak>
  <say-as interpret-as="spell-out">EMI</say-as>
</speak>

<!-- Pause before key information -->
<speak>
  Your balance is <break time="300ms"/>
  <say-as interpret-as="currency">₹5000</say-as>
</speak>

SSML Tag Reference

Problem	SSML Fix	Effect
Currency read incorrectly	<say-as interpret-as="currency">	Reads with correct language context
Digits read as words	<say-as interpret-as="digits">	Forces digit-by-digit: six nine eight seven
Acronym mispronounced	<say-as interpret-as="spell-out">	Spells out letter by letter: E-M-I
Brand name wrong	<sub alias="correct pronunciation">	Replaces text with specified pronunciation
Need a pause	<break time="500ms"/>	Inserts a 0.5-second pause
Speak slower	<prosody rate="slow">	Reduces speech rate for complex information

TTS Reading Newlines from LLM or Agentic Responses

Symptom

When using LLM or agentic app responses in voice flows, the TTS engine reads literal newline characters aloud — saying "backslash n backslash n" instead of pausing. This happens because LLM outputs contain formatting characters (\n, markdown bold, bullet points) that TTS cannot interpret.

Fix

Add a Script node before the Message node to sanitise the LLM output:

// Replace 'YourLLMNode' with your actual node name
var raw = context.steps.YourLLMNode.response || '';

var clean = raw
  .replace(/\\n/g, ' ')         // literal \n strings from LLM
  .replace(/\n/g, ' ')          // actual newline characters
  .replace(/\r/g, ' ')          // carriage returns
  .replace(/\*\*/g, '')         // markdown bold
  .replace(/\*/g, '')           // markdown bullets / italic
  .replace(/#+\s/g, '')         // markdown headers
  .replace(/^\s*[-•]\s*/gm, '') // bullet characters
  .replace(/\s{2,}/g, ' ')      // collapse multiple spaces
  .trim();

context.ttsOutput = clean;
// Reference context.ttsOutput in your Message node

Hindi TTS Failure — Voice-Language Mismatch

Symptom

On language switch to Hindi mid-call, TTS throws a synthAudio requires language error. Logs show language: undefined and voice: undefined. Digits like 6987 may be spoken incorrectly when an English voice receives Devanagari numerals.

Root Cause

The voice name and language code do not match. For example, using language=hi-IN with voice=en-IN-AartiIndicNeural — which is an English-Indian voice, not a Hindi voice.

Fix

Add a Script node before the first Hindi message node:

userSessionUtils.setCallControlParam('ttsLanguage', 'hi-IN');
userSessionUtils.setCallControlParam('voiceName', 'hi-IN-SwaraNeural');
userSessionUtils.setCallControlParam('ttsProvider', 'microsoft');

// CRITICAL: voice name and language MUST match.
// WRONG:   language=hi-IN + voice=en-IN-AartiIndicNeural
// CORRECT: language=hi-IN + voice=hi-IN-SwaraNeural

Supported Hindi Voices

Voice Name	Gender	Language	Notes
hi-IN-SwaraNeural	Female	hi-IN	Recommended — natural Hindi
hi-IN-MadhurNeural	Male	hi-IN	Alternate male voice
en-IN-AartiIndicNeural	Female	en-IN	English-Indian — do not use for Hindi text

Configured Voice Not Applying at Runtime

Symptom

The voice configured in Voice Preferences (for example, hi-IN-SwaraNeural) is not applied during live calls. The bot uses the default English voice instead, with no error shown in the UI.

Resolution

Issue	Resolution
Root Cause	TTS initialises at the first message node. If automation or a script runs first, TTS uses defaults before the voice configuration loads.
Fix 1 (Recommended)	Add a short message node as the very first node in your Experience Flow — before any script or automation node.
Fix 2	After saving the voice config in the UI, edit the flow description field and re-save to force re-registration.
Fix 3 (Script)	Set voice via script: userSessionUtils.setCallControlParam('voiceName', 'hi-IN-SwaraNeural'); before the message node.
Fix 4 (Google)	If you see 'No credentials for Google with labels: undefined' — set ttsLabel using a script, not the UI. See No Credentials for Google.

---

Deepgram ASR / TTS

Utterance Splitting — utteranceEndMs Tuning

Symptom

The user says a long sentence, but Deepgram commits mid-sentence due to natural pauses. Setting deepgramUtteranceEndMs to 3000 or 4000 still results in splits.

Root Cause

deepgramUtteranceEndMs alone is insufficient. It must be paired with deepgramEndpointing to handle micro-pauses within sentences. The endpointing parameter controls Voice Activity Detection (VAD) sensitivity for short gaps. utteranceEndMs controls the final silence window.

Fix

// Recommended for agentic / conversational flows
userSessionUtils.setCallControlParam('deepgramUtteranceEndMs', 1500);
userSessionUtils.setCallControlParam('deepgramEndpointing', 500);
userSessionUtils.setCallControlParam('sttMinConfidence', 0.4);
userSessionUtils.setCallControlParam('deepgramNumerals', true);
userSessionUtils.setCallControlParam('deepgramPunctuate', true);

// For Spanish / Latin American Spanish
userSessionUtils.setCallControlParam('sttLanguage', 'es-419');

Parameter Reference

Parameter	Value / Range	Purpose
deepgramUtteranceEndMs	800–2000 ms	End-of-utterance silence window. Start at 1200; increase by 200 ms if splitting persists.
deepgramEndpointing	300–600 ms	VAD sensitivity. 300 ms = responsive; 600 ms = more patient with pauses.
deepgramNumerals	true/false	Converts spoken numbers to digits (for example, five hundred → 500).
deepgramNer	true/false	Named Entity Recognition — improves proper noun and entity recognition.
deepgramPunctuate	true/false	Adds punctuation to transcripts — useful for NLU processing.
deepgramKeyterms	String array	Boosts recognition of specific listed words.

---

Digit and RX Number Recognition

Symptom

The bot collects multi-digit numbers (for example, 5–7 digit prescription or account numbers). Callers pause between digits. The bot splits the utterance or misses digits.

Production-Tested Configuration

var recognizerConfig = {
  vendor: 'deepgram',
  model: 'nova-3-medical',   // Use nova-3 for non-medical flows
  language: 'en-US',
  interim: true,
  deepgramOptions: {
    endpointing: 300,
    utteranceEndMs: 1000,
    keyterms: [
      'zero', 'one', 'two', 'three', 'four',
      'five', 'six', 'seven', 'eight', 'nine'
    ],
    numerals: true
  }
};

userSessionUtils.setCallControlParam('dtmfBargein', false);
userSessionUtils.setCallControlParam('listenDuringPrompt', true);

Model Selection

Model	Best For	Use Case	Status
nova-3	Best general accuracy	Default IVR flows	Recommended
nova-3-medical	Medical/clinical terms, RX numbers	Healthcare, pharmacy, insurance	Recommended
nova-2-phonecall	Optimised for telephony audio	Call centre deployments	Stable
enhanced	High accuracy, slightly higher latency	Accuracy-critical flows	Stable

---

Google ASR and TTS

16-Digit Card Number — Repeated Digit Failure

Symptom

Google ASR fails on sequences with 8 or more consecutive identical digits (for example, 509099999999990002). The bot may barge in early. This is a known Google ASR limitation — DTMF is the reliable alternative for 16-digit card numbers.

Fix

// Recommended: DTMF input for card numbers
userSessionUtils.setCallControlParam('dtmfCollectMaxDigits', 16);
userSessionUtils.setCallControlParam('dtmfCollectInterDigitTimeoutMS', 4000);
userSessionUtils.setCallControlParam('dtmfCollectTermDigit', '#');

// If speech must be used — enable continuous ASR
userSessionUtils.setCallControlParam('continuousASR', true);
userSessionUtils.setCallControlParam('continuousASRTimeoutInMS', 2500);

// Note: AzureSegmentationSilenceTimeoutMs does NOT affect Google ASR.

Google ASR Model Reference

Model	Description	Use Case	Status
chirp_2	Best for telephony + accents	General IVR	Recommended
chirp_telephony	Optimised for 8 kHz / 16 kHz audio	Telephony deployments	Recommended
telephony	Legacy — stable, widely tested	Default	Good default
telephony_short	Fast, for short utterances (< 5 s)	Menu / Y-N commands	Recommended
long	Batch file processing	Not for real-time IVR	Do not use for IVR

---

No Credentials for Google — Label: undefined Error

Symptom

TTS fails with: No text-to-speech service credentials for Google with labels: undefined. This occurs even after setting ttsLabel in the Voice Preferences UI.

Root Cause

The ttsLabel configured in the Voice Preferences UI does not propagate to the TTS engine at runtime in some configurations. The label must be set using a Script node.

Fix

Add a Script node at the start of your Experience Flow:

// TTS configuration
userSessionUtils.setCallControlParam('ttsProvider', 'google');
userSessionUtils.setCallControlParam('ttsLabel', 'your_google_label');
userSessionUtils.setCallControlParam('ttsLanguage', 'en-US');
userSessionUtils.setCallControlParam('voiceName', 'en-US-Chirp3-HD-Aoede');

// ASR configuration (same approach)
userSessionUtils.setCallControlParam('sttProvider', 'google');
userSessionUtils.setCallControlParam('sttLabel', 'your_google_label');

Background Streaming — Prompts Not Played

Symptom

When using Google TTS with Chirp voices and Background Streaming enabled at the Experience Flow level, follow-up prompts (for example, "Is there anything else?") are logged as synthesised successfully, but no audio is delivered to the caller.

Fix Options

Option	Action
Option 1 — Disable Background Streaming	Go to Experience Flow > Edit > Background Streaming > Disabled. Confirm audio plays correctly.
Option 2 — Use a non-Chirp voice	Switch to a Google Neural voice (for example, en-US-Wavenet-F). Confirmed compatible with Background Streaming.
Option 3 — Use Deepgram TTS	Deepgram TTS has native streaming support and is more stable with Background Streaming enabled.

Sarvam TTS

TTS Fails on Currency, Commas, and Special Characters

Symptom

Sarvam TTS fails on messages containing the ₹ symbol, commas in Indian number format (for example, 3,40,650.23), or mixed punctuation. The Sarvam API does not handle Unicode currency symbols or Indian number formatting natively.

Fix

Pre-process text in a Script node before passing it to Sarvam TTS:

function sanitiseForSarvam(text) {
  return text
    .replace(/₹([\d,]+\.?\d*)/g, function(m, n) {
      var num = parseFloat(n.replace(/,/g, ''));
      if (num >= 100000) return (num / 100000).toFixed(0) + ' lakh rupaye';
      if (num >= 1000)   return (num / 1000).toFixed(0)   + ' hazaar rupaye';
      return num + ' rupaye';
    })
    .replace(/[,;]/g, ' ')       // remove commas and semicolons
    .replace(/\.(?!\d)/g, ' ')   // periods not followed by a digit
    .replace(/[()\[\]{}]/g, '')  // brackets
    .replace(/\s{2,}/g, ' ')     // collapse multiple spaces
    .trim();
}

context.sarvamInput = sanitiseForSarvam(context.llmResponse);

// Provider configuration
userSessionUtils.setCallControlParam('ttsProvider', 'custom:sarvamTTS');
userSessionUtils.setCallControlParam('voiceName', 'simran');

Background Streaming — Audio Not Played

Symptom

When Background Streaming is enabled at the Experience Flow level with Sarvam TTS, audio synthesis is logged as successful in Interactions, but no audio plays to the caller.

Fix

Disable Background Streaming when using Sarvam TTS. Go to Experience Flow > Edit > Background Streaming > Disabled.

Call Flow — Barge-in, Transfer, and Transcription

Aggressive Barge-in — Bot Interrupts User Mid-Speech

Symptom

The bot plays filler music in under 3 seconds, cutting off the user’s utterance. The ActionHookDelayProcessor fires too early, especially when there is a language mismatch between the TTS text and the voice configuration.

Fix

// Disable barge-in for specific nodes (legal, OTP, transfer messages)
userSessionUtils.setCallControlParam('node.bargein', false);

// Delay filler music — prevent premature firing
userSessionUtils.setCallControlParam('actionHookDelayInMs', 8000);

// Session-level barge-in sensitivity
userSessionUtils.setCallControlParam('session.bargein', true);
userSessionUtils.setCallControlParam('session.bargeInSensitivity', 'low');

Scenario Guide

Scenario	Recommended Fix
Bot interrupts greetings	Set actionHookDelayInMs=8000 or higher to give the caller time to state their full intent.
Barge in on legal or OTP prompts	Set node.bargein=false on those specific message nodes.
Background noise triggers barge-in	Set bargeInSensitivity=low. Ensure noise suppression is active in your telephony setup.
Transfer message cut short	Use queueCommand: true — see Transfer Fires Before TTS Message Completes.

---

Transfer Fires Before TTS Message Completes

Symptom

Agent transfer fires before the farewell or transfer message finishes playing. The caller hears a partial message before being transferred.

Root Cause

A direct Agent Transfer Node placed after a Message Node creates a race condition — the transfer may execute before TTS playback completes.

Fix

Use queueCommand: true in a Script node to ensure TTS completes first:

var transferCmd = {
  type: 'command',
  command: 'redirect',
  queueCommand: true,  // Critical — waits for TTS to complete before transfer
  data: [{
    verb: 'transfer',
    destination: 'sip:agent@your-sip-domain.com'
  }]
};

// Execute using your flow's command execution method.>

Transcription Not Displaying — Twilio or ASR Timeout

Symptom

The Transcripts tab is empty after completed calls. This can have two distinct root causes: Twilio webhook misconfiguration or an ASR timeout that is too short.

Diagnosis and Fix

Root Cause	How to Diagnose	Fix
Twilio webhook misconfiguration	The Twilio Media Stream webhook is missing the Kore transcript endpoint. Transcription is enabled, but results are not sent to the platform.	Verify your webhook URL in the Twilio Console includes the correct Kore transcript endpoint for your environment. Contact Kore support to confirm the correct URL.
ASR timeout too short	ASR timeout is set too low (for example, 2000 ms). Calls end with  NO_INPUT before transcription completes.	Increase botNoInputTimeoutMS to 8000 ms or higher. Verify sttMinConfidence is not set too high.
ChannelOverrideTemplate override	Template-level timeout overrides Experience Flow settings silently.	Set botNoInputTimeoutMS explicitly in the ChannelOverrideTemplate itself, not just in the Experience Flow.

---

Quick Reference — All Parameters

All parameters are set using:

userSessionUtils.setCallControlParam('paramName', value);

Parameter	Vendor	Value Range	Purpose
AzureSegmentationSilenceTimeoutMs	Azure ASR	800–2000 ms	End-of-speech silence. Increase to reduce utterance splits.
continuousASR	Azure, Google	true/false	Enable for digit or number collection.
continuousASRTimeoutInMS	Azure, Google	1500–3000 ms	Wait after the last input in continuous mode.
sttMinConfidence	All vendors	0.4–0.7	Minimum confidence to accept an utterance. Lower = more permissive.
sttLanguage	All vendors	BCP-47 code	ASR language (for example, en-US, hi-IN, es-419, ja-JP).
sttProvider	All vendors	Provider string	ASR vendor selection (microsoft, google, deepgram).
botNoInputTimeoutMS	All vendors	5000–15000 ms	Wait before the no-input event. Increase for non-English.
deepgramUtteranceEndMs	Deepgram	800–2000 ms	End-of-utterance silence. Pair with deepgramEndpointing.
deepgramEndpointing	Deepgram	300–600 ms	VAD sensitivity. 300 ms = responsive; 600 ms = patient.
deepgramNumerals	Deepgram	true/false	Spoken numbers → digits (five hundred → 500).
deepgramPunctuate	Deepgram	true/false	Adds punctuation to transcripts.
deepgramKeyterms	Deepgram	String array	Boost recognition of specific words.
ttsLanguage	All TTS	BCP-47 code	TTS language. Must match voiceName language.
ttsProvider	All TTS	Provider string	TTS vendor (microsoft, google, custom:sarvamTTS).
ttsLabel	Google TTS	Label string	Must be set using a script — not UI only.
voiceName	All TTS	Voice identifier	Must match ttsLanguage.
node.bargein	All vendors	true/false	Enable or disable barge-in per node.
session.bargeInSensitivity	All vendors	low/medium/high	Session-level barge-in sensitivity.
actionHookDelayInMs	All vendors	5000–10000 ms	Delay before filler music fires.
dtmfCollectMaxDigits	All vendors	1–20	Maximum DTMF digits to collect (card numbers: 16).
dtmfCollectInterDigitTimeoutMS	All vendors	3000–5000 ms	Wait between DTMF digits.
dtmfCollectTermDigit	All vendors	# or *	Termination digit for DTMF collection.
listenDuringPrompt	All vendors	true/false	Allow ASR to listen while TTS is playing.