3. Viewing and Playing Recordings

After you have found and organized recordings you may view and play the recordings.

3.1. Viewing Recordings

Recordings displayed in the Browse, Search, or Incident tabs are shown in two different ways simultaneously. The first is a simple tabular list displayed in the upper right-hand pane, called the record view. This view can be customized in that it supports several different column and sorting options. The second is a graphical chart of recordings displayed in the lower right-hand pane referred to as the timeline. The timeline can also be customized, but only supports a single method of displaying the recording data by time.

Together, the record view and timeline view provide a comprehensive look at the same set of data. Selections between the two views are synchronized; any recordings selected in one are automatically selected in the other. Similar operations may be performed from each view, including playback and export.

3.2. The Record View

The record view in the upper right-hand pane presents recording data in tabular format. It is also called the call grid. You may sort the information by clicking on the appropriate column heading. A small arrow will indicate the sort direction. To reverse the sort, click the column heading again.

Fig. 3.1 Record View

Columns can also be reorganized. Click and drag any column heading to a new position. The type and number of columns can also be customized. Right click any column heading and a menu of available column names will appear. You may add or delete a column from this menu. To reset the columns to their user-defined default values, select the Reset to Default option.

Fig. 3.2 Record View Showing Records with Removed Media and an Error

There is a retention setting option that allows administrators to retain metadata (duration, date, custom fields, channel number, etc) about record but delete the media (audio or screen recording) whenever the recorder needs to delete recordings to make space. In this case, calls are shown with an [R] in the left-most column of the call grid. A red “!” means there is an error with the call; hover the mouse over the ! to see the error message.

New in version 2024.1.

When editing columns in the Call Grid, a user can simply type the first letter of the required column to quickly search across all options.

Example

After right-clicking a Call Grid column, type the letter “F” to jump directly to all the columns starting with the letter “F”.

Fig. 3.3 Call Grid Quick Search

3.3. Record View Menu Options

Selecting a recording in the record view and then right-clicking on that recording causes the Record View context menu to appear.

Fig. 3.4 Record View Context Menu

The Record View context menu selections include:

• Properties - See Section 5.1: Call Properties

• Block Media Access… - See Section 12.4: Media Access Blocking (Optional, off by default)

• Protect - Checking this box prevents the selected recording from being deleted from the recorder due to recorder call retention policy. When protecting a large number of calls, a progress bar will pop up to provide feedback while the system is protecting calls.

• View Location - View recording’s location on a map if available. See Section {number}: {name}

• New Incident from Selected Records - See Section 4: Creating an Incident

• Add Selected to incident - See Section 4.1: Adding and Removing Recordings from an Incident

• Evaluate Records - Evaluate selected call recordings. Requires Quality Factor add-on license. See Eventide Quality Factor Software Manual for details.

• Start Sequential Playback - See Section 3.9.1: Sequential Playback Mode

• Export… - See Section 6: Exporting Recording Data from an Incident

• Export Single Recording - See Section 6: Exporting Recording Data from an Incident

• Export Using Preset - Exports selected recordings using a saved preset. See Section 6.3: Export Presets and Defaults

• Add Text Annotation - Add annotations to a call, at a specific point in time. See Section 5.3: Annotating Calls

• Add Audio Annotation - Add audio commentary to a call, at a specific point in time. See Section 5.3: Annotating Calls

• Upload User Content - Upload a file to the recorder for recording keeping and use in incidents. See Section 12.2: User Content Uploads

• Show Errors… - This option will only appear if there is an error with the record, indicated by a red exclamation point next to the call. Select it to get see information about the error.

• Generate Speech Factor AI - Enables Speech to Text Integration automatically, or on select playback.

• See Section 3.11: Speech Factor AI Transcriptions

3.4. The Timeline View

The Timeline view in the lower right-hand pane shows the same set of recordings that are displayed in the Record view, but presents the recording data in a linear chart. Recordings are organized into groups by channel name and are displayed in chronological order from left to right. The individual recordings are drawn in the timeline in the horizontal direction in relation to their duration. Scroll the timeline to the right (or to the left) to view recordings that are not visible. If more than one channel is represented in the data, scroll the timeline up or down to display channels that may not be initially visible.

Fig. 3.5 Timeline View - Audio Recordings

You can increase or decrease the time length displayed on the timeline, and increase or decrease the height of the individual rows. To zoom, click on the appropriate small magnifying glass in the lower left corner of the pane. (These are from left to right: vertical zoom in, vertical zoom out, horizontal zoom in, horizontal zoom out.) In Fig. 3.5, it has the words Thirty Seconds in the left corner. Clicking this will open a menu offering a variety of time scales from one second to one year. The selected time scale defines the amount of time to be displayed between each “tick” on the timeline. In Fig. 3.5, there are three ticks, each representing the start of each half minute.

For finer control of the timeline scroll bar:

1. Press and hold your cursor over the scroll bar. The scroll bar will disappear.

2. Use your cursor to navigate through the timeline at a more controlled pace.

Note

If you zoom out on a large amount of call data with several system or text annotations per call, annotations will be displayed as red dots to improve legibility. You will need to zoom in to see details for a specific call.

Fig. 3.6 Example of Annotations as Red Dots at High Zoom

Note

If an Analog board’s channel is configured in the Configuration Manager to record with the detect type On Voxbreak, you can have recordings which are assigned the calltype Inactivity in MediaWorks EXP. They also have a different appearance from regular Audio recordings in the timeline. As shown in Fig. 3.7 below, the timeline contains the standard grey blocks which correspond to Audio recordings as well as transparent blocks which correspond to Inactivity recordings. These Inactivity recordings occur when the audio input levels on the channel fall below the channel’s configured VOX threshold. The difference in appearance makes it easy to distinguish between Inactivity and Audio recordings in the timeline, while the difference in calltype designation makes it easy to search for periods of activity or inactivity. Inactivity recordings can be played from the timeline or the record view table by double-clicking it just as any other call can. For more details regarding “ON VOXBREAK”, refer to the Eventide NexLog Express User Manual.

Fig. 3.7 Example of Inactive and Audio Calls from On Voxbreak Recordings

3.4.1. Adding Bookmarks in the Timeline

The bookmark feature in the timeline allows you to create bookmarks during a 911 call. This feature is especially useful for marking key times for significant events while you continue performing other searches or navigating across different tabs in MediaWorks EXP.

To set a bookmark:

1. Hover your cursor over a position in the timeline and right-click.

2. Select Set Bookmark and choose your bookmark number. You can add up to ten bookmarks to the timeline.

   

   Fig. 3.8 Bookmark Context Menu

Once you add a bookmark, a number will appear at that time.

To navigate back to the bookmark:

1. Press the corresponding number on your keyboard. For example, press 1 to return to Bookmark 1.

   

   Fig. 3.9 Bookmark Icon

Note

Bookmarks are only visible within the date range that was assigned to them. The bookmark will only appear if the selected dates are within the assigned range. If you try to navigate to a bookmark outside the selected dates, you will see the message BOOKMARK OUT OF RANGE.

To remove a bookmark:

1. Right-click the bookmark icon and select Remove Bookmark.

   

   Fig. 3.10 Removing Bookmarks

To clear multiple bookmarks:

1. Right-click anywhere in the timeline and select Clear All.

   

   Fig. 3.11 Clear All Bookmarks Option

3.5. Set Default Timeline Zoom Levels

New in version 2024.1.

This feature can save your default Timeline zoom levels for new tabs so your viewing preferences remain the same for all incidents.

To set the default timeline zoom level in MediaWorks EXP, go to View -> Timeline Default:

Fig. 3.12 Timeline Default

Then select the preferred Timeline Zoom Level:

Fig. 3.13 Timeline Zoom Level

All new Incidents and Search Tabs will now default to the selected Timeline Zoom Level:

Fig. 3.14 Timeline Zoom Level Updated

3.5.1. Playback Settings Menu

In the Channel Name box on the left there are a number of playback settings that you can set on a channel-by-channel basis by using the settings menu.

Fig. 3.15 Playback Settings Menu

They include:

• Volume Slider - Moving the volume slider left or right decreases or increases the volume for this channel. Clicking on the mute button (green speaker icon) to the left of the down arrow mutes the sound on the selected channel. The button will then be grayed out. Clicking on the grayed-out mute button will restore the sound to that channel.

• Balance Slider - Pan this channel to the right, center or left of the stereo field.

• AGC - Checking this box will enable automatic gain control on this channel.

• EQ Boost - Apply an EQ boost or cut based on the EQ settings in the tool menu.

• Options… - Clicking on this menu item displays an Options dialog box. (See Playback Settings Options below.)

3.5.2. Playback Settings Options

The Options dialog box for playback settings allows you to enable Mute, AGC, Obfuscate or EQ Boost for this channel. Mute, AGC and EQ Boost just have an Enable checkbox to turn the feature on and off. The obfuscate option allows you to garble the sound of voices on the channel so that the identity of the speaker is obscured. The Amount slider allows you to adjust the amount the effect is applied to the audio.

Fig. 3.16 Obfuscate Option

3.5.3. Resource Gain Defaults

In the Tools menu, you will find Resource Gain Defaults, which opens to the above panel that allows you to configure a default setting for the EQ, AGC and Gain for every resource name currently on the system, so that every tab you open from then on will load these settings.

Fig. 3.17 Resource Gain Defaults

Reset will return to factory defaults: EQ and AGC Off, Volume at 0db gain. Use Current will change the settings to those set in the currently open tab; this is useful if you’ve gotten it “just right” and want to save the settings for the future. Apply to Tab will apply all of the settings in this pop up to the open tab. Save will save the settings and they will be applied when new tabs are opened.

3.5.4. EQ Settings

In the Tools menu, the EQ Settings menu option opens a pop up that allows you to set the amount of signal boost or cut in a three band EQ. This allows you to tune the playback of call audio to improve sound quality and compensate for lines that are too bright or muddy. One EQ setting is supported at a time; you can either have it turned on for all channels here with Always On setting, or you can turn it on for specific channels only as detailed above.

Fig. 3.18 EQ Settings

3.5.5. Timeline View Menu Options

Selecting a recording in the Timeline view and then right-clicking on that recording causes the Timeline View menu to appear.

The menu selections are the same as those in the Record View menu. Refer to the Record View Menu Options for information about those menu items.

Note: When working in an Incident Tab, there is also a Remove option that can be selected; selecting Remove From Incident allows you to remove the selected recording(s) from the incident.

3.5.6. Timeline Zoom Options

The Timeline has zoom options that let you adjust how much call data you are looking at in a given amount of horizontal space. This ranges in granularity from one second to one year.

Fig. 3.19 Timeline Zoom Options

The Timeline has zoom options that let you adjust how much call data you are looking at in a given amount of horizontal or vertical space. The horizontal ranges in granularity from one second to one year. This grain affects what the distance between two markers on the timeline means. In the example below, the zoom is set to one minute and the dark checkmarks in the timeline are one minute apart:

Fig. 3.20 Timeline At One Minute Zoom

The magnifying glasses with + and - signs can be used to increase or decrease the horizontal (time) zoom level; the arrows change the vertical zoom. Shift+Click into the timeline will zoom in, Shift+Right-Click will zoom out.

3.5.7. Full Timeline Mode

When using the “Search” Tab, MediaWorks EXP will now expand the Timeline view horizontally to offer a larger Timeline window.

• By double clicking the divider between the calendar pane and the callgrid pane, MediaWorks EXP will toggle between Standard View and Full Timeline Mode.

Fig. 3.21 Standard View

Fig. 3.22 Full Timeline Mode

3.5.8. Headless View

When configuring User Settings in NexLog Express, selecting the “Instant Recall” permission allows for a large format view of the most recent calls on the recorder.

Fig. 3.23 Headless View

3.6. Playing Audio Recordings

Once you have located recordings from the Browse, Search, or Incident tabs, you have several options for playing the recordings. The following information describes the options for playing recordings in the Sequential mode. In this mode, recordings are played one recording at a time, in the order in which they appear in the record list. To change the play order, sort the list of recordings differently. Playback will adapt to the new sequence. For information about playing recordings in the Mixed mode, see Playback Modes.

To begin playing the audio from a recording, press the Play button in the Playback console at the bottom of the screen (see The Playback Console). The first recording in the list will begin playing. Toindicate which recording is currently playing a green arrow appears to the left of the recording in the Recording pane, and the call’s row is highlighted in green.

Alternatively, select a recording and press Enter on the keyboard to begin playback, or press the Spacebar on the keyboard to begin, pause, and resume playback. Once the first recording in the list has been selected, the Spacebar can only be used to pause and resume play. To select another recording, highlight the recording in the Record view or in the Timeline view and press the Enter key or double-click on the new recording. The recording will start playing immediately. Use the Spacebar or Pause button to stop playback and the Spacebar or Play button to resume playback.

You may also use the sub-menu items in the Playback menu to control playback or right-click the desired recording and select Start Sequential Playback from the drop-down menu.

3.7. Playing Screen Recordings

Fig. 3.24 Screen Playback

Screen recordings can be played back in the same way as Audio calls, but there are additional options that relate specifically to screens. These options are configured via the Multimedia Display Options window you can access via the View menu:

Fig. 3.25 Multimedia Display Options

This allows you configure how multimedia recordings will be played back, and also allows for configuring the real-time display of Metadata, discussed below. Display in Application is shown above; the screen recording plays back in a floating window inside the MediaWorks EXP browser window. Display in Canvas looks like this:

Fig. 3.26 Canvas at Top

A panel appears above or to the right of the search results (depending on the Multimedia Orientation setting). This panel is called the Multimedia Canvas. It can also appear to the right of the call list:

Fig. 3.27 Canvas at Right

Fig. 3.28 Media Window Controls (Detail)

The top right corner of the media window contains these three buttons: Pop out to Application, Pop Out to External Window, and Close. If popped out to the application level, the first button will show a minus sign for “Collapse into Multimedia Canvas.”

Fig. 3.29 Media Window Popped Out

When popped out to an external window, the multimedia record is still controlled by the Playback Console in the main MediaWorks EXP window. It also just has an X for Close in the upper right corner, which will pop the call back into the main window.

Fig. 3.30 Real-Time Metadata Panel

The Metadata Panel shows a list in chronological order of the metadata associated with calls played back from this search, browse or incident. The line in bold is the current or most recently played call. This lets you highlight the specific metadata you want to keep track of as you play calls.

3.8. The Playback Console

The Playback Console, located at the bottom of the browser window, allows you to control all aspects of playing recordings. From the Playback Console, you can adjust the volume and speed of the recording(s) being played:

• Volume - Use this slider to increase or decrease the master audio volume. Moving the slider to the right increases the volume; moving it to the left decreases the volume. Pressing the 100% button to the right of the slider will reset the slider to the default center position.

• Speed - Use this slider to increase or decrease the playing speed of the recording(s). Moving the slider to the right increases the speed; moving it to the left decreases the speed. Pressing the 100% button to the right of the slider will reset the slider to the default position.

Fig. 3.31 Playback Console

The buttons in the center of the console include (from left to right):

• Previous Recording - If playback is less than 3 seconds into the recording, moves to previous recording in the selected sequence. If playback is more than 3 seconds into the recording, resets to the start of the current recording.

• Play/Pause - Plays any highlighted recording, pauses if currently playing, or resumes playback after pausing.

• Next Recording - Moves to the next recording in the selected sequence.

• Loop - Loops playback for the selected recording or recordings. For sequential playback, this is the beginning and end of a recording or the loop boundary settings within a single recording. For mixed playback this is the loop boundary settings anywhere within the timeline.

To the top of the console are the following additional tools:

• Slider - The slider on top of the console allows you to change the playback position. This is known as scrubbing. You can scrub at any time during playback. To quickly move to a different spot within a recording, click and drag the scrub slider. While scrubbing, keep track of the position by watching the playback time indicator.

• Skip - The double-arrow buttons on each side of the slider allow you to Skip Forward (>>) or Skip Back (<<).

While a recording is being played, its attributes scroll from right to left in the area above the playback controls. This information includes the recording’s channel name, recording time and length of the recording. The information in the area above the slider includes the current playback time. If you click inside this area, it will switch to show the total length of the recording (on the right) as well as current playback position (on the left).

The Playback Console can be displayed in two modes: Static Playback Console (at the bottom of the page) and Floating Playback Console . The mode can be switched at any time from static to float by clicking arrow button in the upper right corner of the Static Playback Console. Once enabled, the Floating Playback Console will be displayed in the upper right corner of the tab by default, but it can be moved around the window. To change the mode back to static, click x close button in the upper right corner of the float console.

Fig. 3.32 Floating Playback Console

The Playback Console state is global for all tabs, except Channels tab and Evaluations tab and saved between the loads. The Static Playback Console is located at the bottom of the browser window. Once started in one tab, the playback can be controlled with the same console in other tabs. Playback Console is displayed as blue in active tab, where the calls are playing from and grey - in other tabs.

Anytime the Playback Console mode can be switched from static to float by clicking arrow button in the upper right corner of the Static Playback Console. Once enabled, Float Playback Console will be displayed in the upper right corner of the tab by default, but it can be moved around the window. To disable Float Playback Console, click x close button in the upper right corner of the float console.

3.9. Playback Modes

MediaWorks EXP supports two modes of playback: sequential and mixed. Recording data can be streamed from an Eventide recorder or from a remote incident containing embedded audio data. Optionally, playback can be modified to include “Time Marker” information at the start of each recording.

3.9.1. Sequential Playback Mode

Sequential playback occurs on one recording at a time. The order in which recordings play is determined by the record view. To change the order in which sequential playback operates, change the sort order of the record view.

To begin sequential playback, select Start Sequential Playback from the right-click context menu in the Record view or Timeline view. Alternatively, you can double-click a recording or highlight a recording and press the Enter key.

3.9.2. Mixed Playback Mode

Mixed playback will play back recordings in the timing that they actually occurred. Mixed Mode playback is also referred to as ‘real time playback’. The order in which recordings plays is determined by the start time of the recording. Mixed playback always proceeds in time order, from earliest to latest. If you are examining recordings from multiple channels, the case may be that multiple recordings occur at the same or overlapping times.

For example, if audio was recorded on a channel named “Dispatch 1” at the same time audio was recorded on a channel named “Emergency 5”, the audio for both recordings will be played back at the same time. When audio is playing in Mixed mode, you can easily follow the progress in the timeline, which highlights all playing recordings across all channels.

3.9.3. Mixed Playback of a Single Resource/Channel

Single timeline resource/channel can be played back in mixed mode. This is useful for playing back a single talkgroup or resource while looking at search results or instant recall on more than one timeline resource at a time. In order to active Single Resource mixed playback hold down the Ctrl key while viewing the properties menu and select “Select Resource Mixed Playback.”

3.9.4. Mixed Playback of Currently Selected Call Records Only

You can playback only a selected subset of calls in the timeline, in mixed mode. Use Ctrl+click to

select multiple calls from the timeline or the call list, and then right click one of the selected calls and choose “Start Mixed Playback.” This will then play in real time all of the selected calls and no other calls.

3.9.5. Quick Playback from Timeline Header: Click to Play

You can click in the timeline header (where the time is displayed along the top) to start mixed playback at any point. We call this “click to play.” If you have started calls in sequential mode, a click to play will follow suit and play in sequential mode. You can switch it back to Mixed by right-clicking in the Timeline header and selecting “Start Mixed Playback.”

3.10. Playback Audit

MediaWorks EXP now offers a played audio indicator to let users see previously played recordings.

Under the “View” dropdown in the menu bar, select “Show Playback Audit”

Fig. 3.33 Show Playback Audit

With this function enabled, the user now has a visible indication of any recordings that have been previously listened to.

Fig. 3.34 Playback Audit

• Grey Dot shows any records listened to in a past session

Fig. 3.35 Past Session Indicator

• Orange Dot shows recordings listened to in current active session

Fig. 3.36 Current Session Indicator

3.11. Speech Factor AI Transcriptions

MediaWorks EXP now features Speech Factor AI Transcriptions, an optional AI-based transcription product that can be configured for NexLog DX-Series recorders. When enabled, this feature offers the ability to transcribe calls as they record, including the ability to record multiple participants during a call. If recordings are separated, AI Transcriptions generates a transcription with each speaker listed individually, providing the ability to generate analytics across all lines.

Speech Factor AI Transcriptions can either run in a hosted cloud environment where encrypted packets summarizing audio are sent to a secure server for analysis, or can be hosted on an on-premises server sold by Eventide Communications.

Speech Factor AI Transcriptions includes the following features:

• Background Transcriptions

• Manual Transcriptions

• ‘Boost’ Transcriptions

Speech Factor AI Transcriptions can be enabled and configured either to run automatically in the background or configured manually. You can transcribe specified channels either in real-time (Background Transcriptions Mode) or can select an individual recording for transcription (Manual Transcriptions Mode) to provide quick and accurate transcriptions of calls and events.

The ‘boost’ transcription feature provides ways to improve the accuracy of AI transcriptions. Each of these transcription modes is covered in detail in the sections that follow.

License Required

Speech Factor AI Transcriptions requires the following Eventide Communications product license for the AI Transcriptions Service.

• DX928 - Speech Factor AI Analytics T1

If your NexLog Express recorder is enabled with this license, you can transcribe recordings in both Background Transcriptions mode and Manual Transcriptions mode.

Note

Background Transcriptions mode and Manual Transcriptions mode were previously known as ‘Always On’ mode and ‘On Demand’ mode. Each mode required a separate license historically. Now only a single license is required, as noted above. Cloud-hosted “Background Transcriptions Mode” recording is sold on an annual basis and is licensed per resource/channel to be transcribed.

3.11.1. Configuring Speech Factor AI Transcriptions

You can configure a NexLog Express recorder for Speech Factor AI Transcriptions. For Speech Factor AI Transcriptions to work, you must first set up a resource group from the web configuration manager of your NexLog DX-Series recorder. For more information, see the NexLog DX-Series User Manual 7.3.4. Resource Groups.

This section covers the configuration of each of the following Speech Factor AI Transcriptions features:

• Background Transcriptions

• Manual Transcriptions

• ‘Boost’ Transcriptions

To configure Speech Factor AI Transcriptions, you must add the license key to the NexLog DX-Series recorder.

3.11.2. Adding the Speech Factor AI Transcriptions License

To add the Speech Factor AI Transcriptions license to the recorder:

1. Sign in to the source NexLog DX-Series recorder’s Configuration Manager with your credentials.

2. Have your 20-digit Speech Factor AI Transcriptions license key ready.

3. In the left navigation menu, select System → License Keys.

   

   Fig. 3.37 Entering the License Key

4. In the right pane, select the Add Key button.

5. In the License Key field, enter your 20-digit Speech Factor AI Transcriptions license key.

6. Confirm that your Speech Factor AI Transcriptions license now appears in the License Key column. It will be listed as ‘Cloud Transcription’. Your license is now active.

   

   Fig. 3.38 License Added

3.11.3. Background Transcriptions

Background Transcriptions refers to the Speech Factor AI Transcriptions feature that can be enabled to automatically transcribe audio calls as they record in real-time. You can read the transcripts as they happen and instantly search for words or phrases.

Note

The number of channels that can be configured for simultaneous transcription in Background Transcriptions mode is determined by the number of channels purchased for the Speech Factor AI Analytics T1 license.

To enable Background transcriptions and apply them to call recordings, you must access the AI –> Transcriptions navigation menu of your NexLog Express recorder’s web configuration manager and enable the Background Transcriptions integration.

To enable background transcriptions:

1. In the left navigation menu, select AI –> Transcriptions. The Transcription page opens.

   

   Fig. 3.39 Enabling Background Transcriptions

2. Make sure that the Integration Enabled checkbox is selected.

3. Under Call Pointer Date/Time, select the calendar icon and then select a date to populate the field.

4. To the right of Call Pointer Date/Time, select the time in hh:mm:sec for each of the three drop-down menu fields to specify when transcriptions will start.

   

   Fig. 3.40 Selecting the Date and Time

5. Under Call Pointer Mode, select Start Time. Call Pointer Mode specifies the date and time at which AI Transcriptions will start transcribing call audio.

6. Channels: specifies which channel(s) will automatically enable transcriptions. You can select an individual channel or a range. However, the preferred method is to select a Resource Group instead to transcribe calls. If you choose this method, you can leave the Channels category blank.

7. From Resource Group: select a Resource Group to use to automatically transcribe calls.

8. In the Languages drop-down menu, choose your language. Speech Factor AI Transcriptions supports multiple languages:

• en-US - English (US)

• es-US - Spanish (US)

• fr-CA - French (CA)

• fr-FR - French (FR)

• de-DE - German (DE)

• ar-AE - Arabic (United Arab Emirates)

• it-IT - Italian (Italy)

You can add up to four languages, which are then listed on-screen. The default language is “English (United States)”.

Note

Additional supported languages include Russian and Japanese. To request more language code options, contact service@eventidecommunications.com for more details.

9. Service Region: specify the processing region for transcriptions.

Fig. 3.41 Selecting Resource Group, Language, and Service Region

Once you have configured Background Transcription mode, MediaWorks DX will now automatically generate Speech Factor AI Transcriptions on all enabled resources in the background.

3.11.4. Manual Transcriptions

You can also configure transcriptions manually. Just as in Background Transcriptions mode, in Manual Transcriptions mode, you can generate manual AI transcriptions on any channel or recording by configuring and enabling the Manual Transcriptions integration from your NexLog Express recorder’s Web Configuration Manager, under AI -> Transcriptions.

To generate manual transcriptions:

1. In the MediaWorks DX user interface, select a call.

2. Right-select the call and choose Generate Transcription.

   

   Fig. 3.42 Generating Manual Transcriptions

   The call transcription is now generated and displayed in the Call Properties dialog on the Transcription tab.

   

   Fig. 3.43 Manual Transcriptions

3. As before, from your MediaWorks DX user interface, right-select a recording and then select Generate Transcription.

   Note

   Manual Transcriptions mode is billed by the number of hours of recordings transcribed, available in quantities of 1000 hour increments.

4. Sign in to your MediaWorks DX user interface.

5. Right-select a recording and then select Generate Transcription.

Once you have configured your choice of Transcription mode, MediaWorks DX will now automatically generate Speech Factor AI Transcriptions on all enabled channels in the background. You can now search for and view your transcriptions in MediaWorks DX.

See Searching for Transcriptions in MediaWorks DX for more information.

3.11.5. Configuring Boost

The ‘Boost’ feature provides ways to correct words or multi-word phrases that AI may mis-transcribe. Mis-transcriptions may occur either because AI interprets a word or phrase as very similar to one found in its AI model, or because it is not found in it.

‘Boost’ increases the likelihood that AI will correctly transcribe a word or phrase. The best strategy to use to have AI correct its transcription depends on whether the word or phrase is found in AI’s model. Here, trial-and-error is required because there is no way to know this in advance.

3.11.5.1. Before Configuring Boost

Before configuring Boost for a customer, Eventide Communications suggests monitoring how transcription is working practically in the real world, by looking at 911 calls. Note that Boost is only required if issues exist to be resolved. As an important example, is the Agency Name being correctly transcribed? Is there a specific pattern of mis-transcribed words or phrases? If not, configuring Boost may not be needed currently.

If a specific transcription problem(s) exists, set up Boost, then monitor again.

3.11.6. Using Boost

The ‘Boost’ feature is best explained by presenting some practical use cases. Three practical real-world examples now follow. The first two examples assume that a word or phrase is found in AI’s model.

Example 1:

Suppose you have two similar or identical-sounding valid words, such as ‘whether’ and ‘weather’, and that both words are found in AI’s model. If you wanted the word ‘whether’ to be the more frequently-transcribed word, you would assign it a greater weight using Boost.

To boost a word or phrase:

1. In the Boost Phrase field, enter the desired word, ‘whether’.

2. In the Boost Value field to the right, from the drop-down menu, select the highest weight value ‘20 (Very Likely)’.

3. To the immediate right in the Boost Value field, from the following drop-down menu options, choose ‘20 (Very likely)’.

   • 0 (Very Unlikely)

   • 5 (Unlikely)

   • 15 (Likely)

   • 20 (Very likely)

4. Select the Add Boost button.

   

   Fig. 3.44 Boosting a Word or Phrase for Transcriptions

Following this procedure maximizes the probability that AI will choose “whether” over “weather”, the majority of the time.

Example 2:

Similarly, suppose your agency name is ‘Barton County’, but AI mis-transcribes it as ‘Pardon County’. The assumption is that this is because although the phrase ‘Barton County’ is found in AI’s model, it is not at the top of the list of AI’s phrases. In this case, you can boost ‘Barton County’ so that it is correctly transcribed.

1. In the Boost Phrase field, enter the (correct) phrase, ‘Barton County’.

2. In the Boost Value field, select the greatest value “20 (Very Likely)”.

3. Select the Add Boost button. ‘Barton County’ is now likely to be transcribed the majority of the time.

   

   Fig. 3.45 Configuring the Boost Value Field

Similarly, add additional words or phrases as per customer requirement. Service Provider technicians should boost the end customer’s Agency Name. Each added boost phrase becomes listed on-screen with its corresponding boost value.

Conversely, had you assigned the lowest weight of “0 Very Unlikely” to ‘Barton County’, this would minimize the likelihood of ‘Barton County’ being transcribed, as opposed to similar-sounding phrases in AI’s model. If this strategy fails, the word or phrase will likely not be found in AI’s model.

Example 3:

In this example, the assumption is that a mis-transcribed word or phrase is not in the AI’s model. This scenario requires a different strategy consisting of just two steps. Continuing with the “Barton County” example, again, suppose that phrase is mis-transcribed as ‘Pardon County’. In this case, ‘Barton County’ is unknown to AI. To remedy this consists of a two-step procedure.

In step one, what we can do is boost ‘Pardon County’— the mis-transcribed word —which is in AI’s model. (The reason for this will be made clear in step two). This gives ‘Pardon County’ the highest weight among similar phrases known to AI. The result is to increase the likelihood that the AI will transcribe that phrase.

In step two, we can use Search and Replace to create a fixed, one-to-one mapping between the mis-transcribed phrase, ‘Pardon County’, and ‘Barton County’, so that AI is now highly likely to transcribe ‘Barton County’ over any other phrase the greatest percentage of the time. The following procedure illustrate this.

To map a boosted word or phrase to another using find and replace:

1. In the Boost Phrase field, enter ‘Pardon County’, the mis-transcribed phrase.

2. In the Boost Phrase Value field, from the drop-down menu select “20 (Very likely)”.

   

   Fig. 3.46 Boosting a Word or Phrase Using Find and Replace

3. Select Add Boost.

4. In the Find field, enter ‘Pardon County’.

5. In the Replace field, enter ‘Barton County’.

6. Select the Add Mapping button. This mapping now maximizes the likelihood that AI will correctly transcribe ‘Barton County’ instead of ‘Pardon County’.

7. Repeat this process to configure replacement words for unknown names, adding word or phrase mappings as required. Each new mapped pair will appear on-screen across the page.

8. Leave the Service Account Key section as-is. The Service Account Key will have been uploaded earlier already by an Eventide Communications technician, as indicated by the green check mark in the next screenshot.

9. By default, only configure the ‘911 Positions’ Resource Group for transcriptions.

   

   Fig. 3.47 Service Account Key

10. Select Save.

3.11.7. Searching for Transcriptions in MediaWorks DX

You can search for, and view your transcriptions in MediaWorks DX using the Transcription Tab search function. Transcription Tab Search is a robust search filter option that gives you the ability to find keywords directly from call transcriptions.

To search for transcriptions:

1. Sign in to MediaWorks DX with your credentials.

2. Select the Search icon.

   

   Fig. 3.48 Select Search Icon

   The AI Transcriptions Search Tab dialog appears.

   

   Fig. 3.49 Transcriptions Tab

3. In the Transcription tab Search window, type a search term in the search field and then select Go. All available transcription records that contain your search term will be listed in the Transcription tab Search window.

4. Select a target call from the results to listen to it.

5. To view the transcription, hover your mouse at the bottom of the timeline over the thin grey bar. The transcription text now becomes visible.

6. To confirm the accuracy of transcription results, listen to the call audio by selecting a call listed in the Transcription Tab Search window. Each call is shown with a green right-pointing arrow.

   

   Fig. 3.50 Transcription Tab Search

You can view Transcriptions in recordings by hovering your mouse towards the bottom of the timeline, on its grey section, and also within the recording’s call properties.

3.11.8. External Media

Speech Factor AI Transcriptions allow for external media such as voice, text, photos, and videos to be uploaded to the recorder for transcription, analysis, and exporting.

3.11.9. Radio Transcription

Speech Factor AI supports radio transcription. Quality of transcription may vary based on the input source.