Widget Voice

There are two core functionalities available in the multi-agent Widget Voice feature.

Table: Two Functionality of Voice

Configuring Widget Voice

After clicking Enable Agent Voice under the Voice menu, you will access the Widget Voice configuration panel to fine-tune chat interaction settings:

• Voice Configuration: Choose between End-to-End Voice (direct voice-to-voice format) or STT / TTS (Speech-to-Text and Text-to-Speech format).
• Speech to Text Model: Select the model used for converting user speech text (e.g., nova-2-general).
• Text to Speech Model & Voice: Choose your preferred text-to-speech model and specific voice style (e.g., Zain - Confident, Bright and Fresh).
• VAD Threshold: Adjust voice activity detection sensitivity (0.1 to 0.9) to handle quiet or strong speech signals.
• Silence VAD: Set the duration of silence (in seconds) before treating speech as finished.
• Filler Sound Duration: Define how long filler sounds play (in seconds) during processing.

3.1. Integrating and Configuring AI Models for Widget Voice

To add and configure an AI model for your platform:

1. Access Settings: Navigate to the home page and click Account Settings from the left-hand navigation panel.
2. Open Integrations: Go to the Integrations tab and select the AI Models sub-menu to view the available provider cards (e.g., Google Gemini, OpenAI).
3. Select Provider: Choose your desired AI provider card and click Manage.
4. Create Dynamic Provider: Click the + Create button to open the configuration modal, enter an Integration Name, and click Fetch to load available models.
5. Enable Models: Browse the list of available models, toggle on the switches for the models you want to add, and click Create Provider.
6. Configure Product Access: Edit your configured model and select the checkboxes under Product Access (such as Realtime Voice, Speech to Text, or Text to Speech) to enable your desired voice features, then click Save.

3.2. Default Agent Voice

Table: Agent Voice of RealTime Voice

3.3. Voice Tone Instruction

The voice tone instruction can be added to control and customize the voice tone of Multi Agent.

4. Campaign Voice

Under the Voice menu, users can configure two distinct voice features:

• Agent Voice Interaction: Enables voice input within the chat widget for seamless interaction with agents.
• Campaign Telephone Voice: Enables inbound telephone voice support, allowing callers to reach the agent chain via Twilio phone support.

4.1. Campaign Voice Configuration

Use this section to configure inbound telephone calls and behavior for your campaign voice agent.

• Voice Configuration: Choose between End-to-End Voice (direct voice-to-voice conversation) or STT / TTS (Speech-to-Text and Text-to-Speech format).
• Inbound Phone Number: Enter the optional Twilio phone number (in E.164 format) that customers dial for inbound calls. Leave blank for outbound-only campaigns.
• Model: Select the AI model used for the voice conversation.
• Voice: Choose the sound of the agent's voice during the conversation (e.g., Alloy).
• Voice Speed: Adjust the agent's talking speed (default is 1.0 — Normal).
• VAD Threshold: Set the Voice Activity Detection sensitivity (0.1 to 0.9). Lower values detect quieter speech; higher values require stronger speech signals.
• Silence VAD: Define the silence duration (in seconds, ranging from 0.1 to 3) before treating speech as finished.
• Agent speaks first: Toggle on to have the agent start speaking immediately when the call begins, or toggle off to let the user speak first.

• Twilio Webhook URL (optional): Copy and paste this URL into your Twilio Console under your phone number settings (go to Phone Numbers > Manage > Active numbers, select your number, set "A call comes in" to Webhook (HTTP POST), paste the URL, and save). This connects inbound calls to your agent chain.
• Disable Campaign Telephone Voice: Click this button at the bottom if you want to turn off and reset the voice feature for this agent chain.

4.2.1. Creating a New Campaign

To set up an outbound voice campaign:

• Navigate to the Campaign tab within your agent chain.
• Click the + Create button to open the campaign configuration panel.
• Name: Enter a descriptive name for the campaign.
• Twilio Phone Number: Provide the phone number used to reach out to recipients.
• Click Continue to proceed or Cancel to abort.

4.2.2. Adding Campaign Recipients

You can add recipients manually or import them via spreadsheet:

• Manual Entry: Click + Add to open the recipient form. Enter the recipient's Phone number and optional Call context fields (Title/Value pairs) to provide extra context for the agent, then click Add.
• Bulk Import: Click Import CSV / Excel to upload multiple recipients at once.

4.2.4. Proceeding to Call Settings

Once you have entered the campaign name, Twilio phone number, and added your recipients:

• Click the Continue button to advance to the agent's on-call behavior settings.
• Click Cancel to discard your changes.

Configuring Campaign Settings and Creating the Campaign

In the final setup window, configure how the agent interacts during calls:

• Note: Any configurations made here will automatically apply to your campaign voice management in the general settings.
• Voice Configuration: Choose between End-to-End Voice (voice-to-voice conversation) or STT / TTS (Speech-to-Text and Text-to-Speech format).
• Model: Select the underlying model for the voice conversation.
• Voice: Select the voice sound for the agent (e.g., Alloy).
• Voice Speed: Adjust the agent's talking speed.
• VAD Threshold & Silence VAD: Fine-tune speech detection sensitivity and silence duration before speech is treated as finished.
• Click Create Campaign to finalize and launch your campaign, or click Back to return to recipient setup.

4.2.5. Campaign Overview and Execution

After finalizing your campaign creation, you will land on the campaign dashboard to review and execute your setup:

• Overview Metrics: Monitor real-time performance indicators including Total Attempt, Completed, and Failed calls.
• Campaign Details: View core configuration data such as the campaign Name, Twilio Phone Number, Total Recipients, and creation timestamp (Created At).
• Editing Settings: Click Edit Campaign in the top-right corner to modify details if needed.
• Running the Campaign: Click the Run Campaign button at the bottom of the page to initiate the voice campaign execution.

4.2.6. Reviewing Campaign Execution and Call Details

After running a campaign, you can track the outcome and review individual interaction records:

• Viewing Completed Runs: Locate your executed campaign run under the Results section (marked with a Completed status badge). Click on the run entry to open its detailed breakdown.
• Checking Call Script: In the detailed view, select the Call script tab to review the conversation log or transcript for specific recipients.
• Inspecting Call Details & Duration: Click the Details tab next to the call script to view metadata such as call duration, attempt timestamps, and status information.