The Voice Fingerprints page is your centralized workspace for enrolling, managing, and monitoring employee voice profiles used by Call AI. These voiceprints enable accurate speaker identification across all calls—even when names are not stated—by analyzing an employee’s unique vocal characteristics.
From this page, you can:
Upload or record new voice samples
Retrain the AI model to improve identification accuracy
View processing status and confidence scores for each employee
Location: Go to New Settings > Voice > Voice Fingerprints
Navigation and Layout Overview
Top Section Controls
Search & Filter Bar
Filter employees by fingerprinting status:All
Ready (fully enrolled)
Training (in progress)
Pending (submitted but not yet processed)
Failed (processing failed)
Create Voice Fingerprint
Launches the fingerprint enrollment interface for a selected employee.How It Works
Opens a modal explaining best practices for creating voice fingerprints.Refresh Button
Manually refreshes fingerprint processing statuses and confidence scores.
Employee Sections
Each employee has a collapsible section with key fingerprinting information:
Name and Status
Displays the employee’s name and fingerprint status. On the right, you’ll see:
Fingerprint Status – e.g., Ready when fingerprinting is complete
Confidence Score – e.g., 96.0%, indicating match reliability
Voice Sample Summary
A summary line shows how many samples are processed (e.g., 3/3 processed). Click the dropdown (▾) to view individual sample details.
Voice Sample Details (Expanded View)
When you expand an employee section, you’ll see a list of voice samples with the following fields:
Column | Description |
Name | The filename of the audio clip typically indicates the channel (e.g., 230354_right.mp3). |
Status | Indicates if the sample is READY, meaning it has been successfully processed. |
Quality | A green checkmark means the sample meets quality standards for clarity, signal-to-noise ratio, and duration. |
Created | The timestamp when the sample was added (useful for audit trails). |
Type | The source or reason for the sample. Most are labeled enrollment, meaning they were added intentionally to build the fingerprint. |
Actions | Playback and delete icons: |
Play the sample
Delete the sample (not shown in the screenshot but typically available via action buttons)
Adding Voice Samples
To further enhance fingerprint accuracy, click the microphone icon to upload more samples.
Tooltip: Add voice samples to improve fingerprint accuracy
Upload Options
Record Audio
Choose from:
English – Verification Paragraphs
English – Natural Speech
Español – Verificación De Voz
Select and record at least 2 samples.
Upload Files (via the second tab)
The more high-quality and varied the samples, the better the match accuracy in future calls.
Retraining the Model
Once samples are uploaded, you can retrain the AI fingerprinting model to boost accuracy using the brain icon.
Tooltip: Retrain model to potentially improve accuracy
Retraining may improve model accuracy with existing samples.
Use this when:
Fingerprints are already “Ready” but performing inconsistently.
You’ve added more voice samples and want the model to reprocess them.
How Confidence Score Works
The Confidence Score (e.g., 96.0%) indicates how reliably the system can match a speaker to their fingerprint. Scores are influenced by:
Audio quality (clarity, minimal background noise)
Sample quantity (3–5 is ideal)
Variety of tone and speech content
Best Practices
Enroll All Call-Handling Staff: If you enroll one employee, you must enroll all employees who handle calls. Partial enrollment leads to inconsistent call attribution and weakens overall accuracy.
Monitor Confidence Scores: Add more samples for employees scoring below 90%.
Use “Ready” as a Benchmark: Employees not marked Ready should be prioritized for follow-up.
Keep Fingerprints Current: Remove or reassign fingerprints if employees leave the practice or change roles.
Use the Voice Fingerprints Page When:
You're onboarding a new employee who handles calls
You want to improve system-wide speaker recognition
You need to manage or delete outdated voiceprints
You’re aiming to increase the Confidence Score for a specific person