---
title: "Speaker Diarizer"
slug: "dataflow-nodes-speaker-diarizer"
updated: 2026-02-10T04:34:24Z
published: 2026-02-10T04:34:24Z
canonical: "docs.knovvu.com/dataflow-nodes-speaker-diarizer"
---

> ## Documentation Index
> Fetch the complete documentation index at: https://docs.knovvu.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Speaker Diarizer

Takes a single channel(mono) audio in which more than one people talks and splits this into a multichannel audio where each channel only has a single person's speech. 

## Parameters
| <div style="width:170px">name</div> | description | <div style="width:120px">default</div> |
|---|---|---|
| NumSpeakers | Expected number of different speakers in the input audio. The most common use is with records from call centers. So, the default value is 2.   | 2 |
| MaxNumSpeakers| Expected number of different speakers in the input audio. -1 for no upper limit. | -1 |



## Inputs

### Audio
&emsp;Accepts audio from a single channel. 


### Events
&emsp;none

## Outputs

### Audio
&emsp;Outputs audio to multiple channels

### Events
| name | description |
|---|---|
| [Start Of Segment](/v1/docs/events#start-of-segment)| Raised once before sending each audio fragment. This event is sent to the same channel with the corresponding audio. |
| [End Of Segment](/v1/docs/events#end-of-segment) |  Raised once after sending each audio fragment. This event is sent to the same channel with the corresponding audio. |

## Remarks  
Speaker Diarizer splits the provided audio into many smaller fragments and determines the speaker in each fragment. Sends each speaker's audio on a separate channel output. 

Speaker Diarizer technology needs to analyze the whole audio file at once. So, this node can only work in batch mode and cannot be used for streaming.

:::(Info) (Important Note)
Speaker Diarizer splits the audio into separate segments. So, there is no need to use another [Audio Splitter](/v1/docs/dataflow-nodes-audio-splitters){target=`_blank`} node in the project.
:::

You can adjust the number of output channels from the menu that appears by right clicking the node. 
By right clicking the node, we can add and remove channels from one channel to three channel outputs. Each output sends the audio of only one speaker. 

### Project Structure
A minimal project that utilizes Speaker Diarizer can be created as such:
<center>
    
![image.png](https://cdn.document360.io/9bca2910-6a3b-4224-9470-43f91f9a6d57/Images/Documentation/image%28901%29.png){height="" width=""}
</center>

Note that in this project, the [SR Http](/v1/docs/dataflow-nodes-speech-recognition-sr-http){target=`_blank`} results will be tagged as `Speaker No: 0` and `Speaker No: 1`. 

### Supported flow types
Batch

## Release Notes

<div style="max-height: 200px; overflow-y: auto; overflow-x: hidden">
     <details>
        <summary> <strong> v4.5.0</strong> </summary>
        <ul>
            <li>Removed 'Address' and 'Ignore SSL Errors' from parameters. </li>
        </ul>
    </details>
    <details>
        <summary> <strong> v1.0.0 </strong> </summary>
        <ul>
            <li> Introduced Node.</li>
        </ul>
    </details>
</div>