Theme
Voice Analytics
The Voice Analytics dashboard provides metrics specific to voice channel conversations. It covers call quality, IVR navigation patterns, speech recognition performance, and caller sentiment -- giving you a complete picture of how your voice bots are performing.
Accessing Voice Analytics
Navigate to Monitoring > Analytics > Voice Analytics. The dashboard is available if you have at least one bot deployed to a voice channel. If no voice data exists for the selected period, the dashboard displays a prompt to configure a voice channel.
KPI Cards
The top of the page displays headline metrics for your voice operations.
| Metric | Description |
|---|---|
| Total Calls | Number of voice calls handled by your bots in the selected period |
| Avg Call Duration | Mean duration from call answer to call end |
| IVR Completion Rate | Percentage of callers who navigated the full IVR menu without abandoning |
| Transcription Accuracy | Estimated accuracy of speech-to-text transcriptions based on confidence scores |
| Containment Rate | Percentage of calls resolved by the bot without agent escalation |
Voice analytics dashboard with KPI cards for total calls, average call duration, IVR completion rate, transcription accuracy, and containment rate, plus call duration histogram
TIP
Low IVR completion rates often indicate confusing menu options or too many levels. Review the IVR path analysis to identify where callers drop off.
Call Duration Analysis
A histogram shows the distribution of call durations across the selected period. This helps you identify typical call lengths and detect outliers.
| Segment | Description |
|---|---|
| Under 30 seconds | Likely abandoned or quick self-service lookups |
| 30 seconds -- 2 minutes | Typical automated resolution range |
| 2 -- 5 minutes | More complex interactions, possibly involving multiple IVR steps or tool calls |
| Over 5 minutes | May indicate bot confusion, loops, or delayed escalation |
IVR and DTMF Usage
If your bots use IVR Menu nodes, this section tracks how callers interact with touch-tone menus.
DTMF Key Distribution
A bar chart shows how often each DTMF key (0-9, *, #) was pressed across all IVR menus. This reveals which menu options are most popular and which are rarely used.
IVR Path Analysis
A flow diagram visualizes the most common paths callers take through your IVR menus. Each step shows the percentage of callers who chose that option. Dead-end paths where callers abandon or request an agent are highlighted.
No-Match Rate
The percentage of IVR interactions where the caller's input did not match any configured option. A high no-match rate suggests that callers expect options that are not offered or that speech recognition is misinterpreting spoken responses.
TTS and STT Performance
These metrics track the quality of text-to-speech and speech-to-text processing.
| Metric | Description | Target |
|---|---|---|
| STT Confidence Score | Average confidence of speech recognition across all utterances | Above 0.85 |
| STT Latency | Mean time from end of speech to transcription result | Under 500 ms |
| TTS Latency | Mean time from text submission to audio playback start | Under 300 ms |
| Fallback to DTMF | Percentage of interactions that fell back from voice to touch-tone input | Under 10% |
WARNING
STT confidence scores below 0.80 may indicate issues with audio quality, caller accents, or domain-specific vocabulary. Consider adding custom vocabulary to your speech recognition configuration.
Call Recording Playback
For calls where recording is enabled, you can listen to the audio directly from the analytics view.
- Navigate to the call list within Voice Analytics.
- Click the play icon next to any call to open the audio player.
- The player displays the waveform alongside the synchronized transcript, highlighting each utterance as it plays.
Recordings are stored according to your workspace's data retention policy.
Call recording playback interface showing audio waveform visualization with synchronized transcript highlighting, play/pause controls, and timeline scrubber
Sentiment Analysis
OmniBots analyzes caller sentiment throughout each voice conversation. Sentiment is scored on a scale from negative (-1.0) to positive (1.0) and tracked at each turn of the conversation.
| Indicator | Description |
|---|---|
| Overall Sentiment | Average sentiment score across all calls in the period |
| Sentiment Trend | Line chart showing how average sentiment changes over time |
| Negative Call Rate | Percentage of calls with an overall negative sentiment score |
| Sentiment by Outcome | Comparison of sentiment for contained vs escalated calls |
Calls with strongly negative sentiment are flagged in the call list for review.
Filtering and Export
Filter voice analytics by bot, date range, call outcome (contained, escalated, abandoned), or minimum call duration. Export the filtered data as CSV or PDF using the Export button.
Next Steps
- Tool Analytics -- Monitor tool execution performance
- Custom Reports -- Build reports combining voice metrics with other data
- Conversations -- Review full transcripts of voice conversations