Speech-to-Text (STT)

STT is responsible for:

• Converting caller audio into transcript text

• Detecting end-of-turn pauses

• Supporting language detection

• Providing confidence signals

Transcript accuracy directly affects:

• Intent identification

• Question sequencing

• Location capture

• Summary quality

Depending on deployment, configurable elements may include:

• Language model selection

• End-of-turn sensitivity

• Smart formatting options

• Confidence thresholds

Improper STT configuration can lead to:

• Misidentified intent

• Excessive clarification

• Conversation instability

1. Place controlled test call.

2. Speak clearly structured data (address, phone number).

3. Review transcript in Triage.

4. Confirm:

   • Numbers formatted correctly

   • No missing words

   • No repeated fragments

5. Compare transcript against actual speech.

Frequent transcript errors may indicate provider or tuning issues.