Search Text in Calls

The Search Text in Calls page allows users to locate relevant call insights efficiently using advanced speech-to-text search capabilities.

To open this functionality, go to Settings > Speech Recognition Management, then switch to the Search Text in Calls tab. Alternatively, use the Navigation Assistant in the Dialer or Cloud PBX.

Using this page, you can:

  • Utilize the date filters to refine the list of calls
  • Indicate keywords or phrases to perform a search in the call transcript database.
  • Examine the generated call summaries to locate key discussion points promptly.
  • Sort the results by call rank to focus the analysis on the most impactful interactions.

Search Filters

The Search call by filters functionality provides the following filtering options to enhance precise modification of search results.

Screenshot of the Speech Adapter section

Speech Adapter section

  • Choose the appropriate option from the Speech Adapter dropdown menu. The available adapters are based on the current configuration of the Speech Recognition Channels.
  • Language: Select a particular language from the language drop-down menu to filter calls accordingly. Alternatively, retain the [Any Language] selection to search the entire language set.
  • Date: Optionally, set a date range using the From Date and Till Date fields.
Screenshot of the prase, time and summary filters

Phrase, Time and Summary filters

  • Phrase: Locate occurrences within the call transcripts by entering a keyword or phrase. Then you can find calls with the designated text in the speech-to-text transcription.
  • Phrase Start Time & Phrase End Time: To narrow the search within the call transcript, set a phrase start time and phrase end time (in seconds). The system will locate the specified phrase only if it occurs between these times.
    For example, entering three as the start and 10 as the end will restrict the search to the interval between the 3rd and 10th seconds of the call.
  • Summary: The system utilizes an AI tool that transcribes calls and generates concise summaries highlighting key subjects and dialogues.
    You can enter keywords to search within these summaries and quickly find relevant calls without reviewing full transcripts.
Screenshot of the rank settings

Rank settings

  • Speech Recognition Rank: AI analyzes calls and generates rationales explaining the assigned quality score. You can search specific words or phrases within these descriptions to quickly find calls based on the quality score explanation. The full Rank Description is available under the Transcription column in the search results table.
  • Minimum & Maximum rank: You can set minimum and maximum rank values to filter calls by their quality score and control which calls appear in the search results.

Search Output Columns

The table below provides descriptions of the output columns.

ColumnDescription
IDThe unique identifier of the call.
Start AtThe timestamp indicates when the call was initiated
Bill DurationDuration after the call was answered
Caller IDThe agent’s phone number that was used to make the call
ToThe destination phone number
ChannelsThe number of audio channels recognized during speech analysis. Typically 2 for calls involving both a client and an agent.
LanguageThe detected language used during the call conversation
RankThe quality score assigned to the call based on AI analysis
TranscriptionYou can click on Show Transcription to view the full call transcript, call summary & rank description
RecordingAudio recording of the call

Transcription & Recording

Screenshot of Call Records with the Show Transcription button highlighted

Call Records with the Show Transcription button highlighted

When you click Show Transcription, a pop-up window opens, allowing the user to review call data, analyze conversations, and retrieve information without leaving the current view.

The window is divided into three main sections:

  • Summary: Displays the AI-generated Call summary and the Rank Description, alongside the assigned rank score.
Screenshot of the window with the call transcription controls

Window with the call transcription controls

  • Transcription: Displays the whole AI-transcribed conversation. Each entry includes a timestamp and speaker identification.
    You can select the number of entries to display per page (10, 25, 50, or 100). Additional options allow you to search for specific words or download the transcript as CSV or JSON files.
Screenshot of the AI-transcribed conversation

AI-transcribed conversation

  • Call details: Displays essential call information, including the call ID, bill duration, destination number, and Caller ID.

    Screenshot of call details

    Call details

Additionally, you can click Play Record to open the audio player with a waveform view to quickly identify spikes, pauses, and transitions. Playback controls include play, pause, rewind, fast-forward, volume adjustment, mute, channel selection, and speed control (0.5×–2×). The recording can be downloaded locally.

To view additional call details, click on Extended Data to open the whole Show Transcription section, which includes the call summary, rank description, and complete searchable transcript with download options.

Screenshot of the whole Show Transcription section

The whole Show Transcription section