Multi-speaker

The Multi-speaker feature uses real-time speaker diarisation to automatically identify and distinguish different voices speaking into the microphone. It is specifically designed for multi-party interactions — such as client consultations, contract negotiations, depositions, witness interviews, and team meetings.

When enabled, SpeechLive Direct transcribes the conversation in real time, attributing each utterance to its respective speaker with an exact elapsed timestamp.

[Note]Note

  • Multi-speaker is located under the Speech recognition section in the Main menu of the SpeechLive Desktop App.

  • Enabling Multi-speaker automatically engages Auto punctuation and locks it in the On position.

Enable Multi-speaker Mode

  1. In the SpeechLive Desktop App, click the Main menu icon (☰) in the top-left corner.

  2. Click Speech recognition to expand the section.

  3. Locate the Multi-speaker setting.

  4. Toggle the switch to On.

    Notice that the Auto punctuation toggle immediately turns On and is locked in place.

  5. Position your microphone centrally so all participants can be clearly captured.

  6. Begin the conversation. The transcript will separate utterances by speaker in real time.

  7. To return to standard individual dictation, toggle Multi-speaker back to Off.

Transcript Output Format

During a Multi-speaker recording session, SpeechLive Direct formats each speaker turn into a structured, chronological entry: [HH:MM:SS] Speaker X: <punctuated text>

  • [HH:MM:SS] — Elapsed time from the start of the dictation session (hours, minutes, seconds).

  • Speaker X: — Identified speaker label assigned to each detected voice (e.g. Speaker 1:, Speaker 2:).

  • <punctuated text> - Transcribed speech, automatically punctuated for readability.

Automatic Link with Auto Punctuation

Because participants in a live conversation, meeting, or interview speak naturally and do not dictate verbal punctuation commands (such as "comma" or "period"), Multi-speaker is functionally linked with Auto punctuation:

  • Automatic activation: When you turn the Multi-speaker toggle On,Auto punctuation is turned On automatically.

  • Locked setting: The Auto punctuation switch is locked in the On state and cannot be disabled while Multi-speaker is active.

  • Unlocking: When you turn Multi-speaker back to Off, the Auto punctuation toggle is unlocked, allowing you to choose whether to use automatic or manual punctuation for single-speaker dictation.

Best Practices for Multi-speaker Sessions

Table 24. Best Practices for Multi-speaker Sessions

Practice

Recommendation

Microphone type & pickup pattern

Use an omnidirectional microphone: Avoid narrow-directional (unidirectional or close-talk headset) microphones designed for individual dictation, as they suppress off-axis voices. Use an omnidirectional or conference/boundary microphone capable of capturing 360° ambient room sound so all participants around the table are picked up clearly and evenly.

Disable noise cancellation

Turn off noise cancellation: Disable active noise cancellation (ANC) and background noise suppression features on your microphone hardware, driver, or operating system settings. Aggressive noise cancellation algorithms often misidentify secondary speakers situated further from the device as ambient noise and filter them out.

Microphone placement

Place the microphone centrally on the table, equidistant from all participants, ensuring an unobstructed sound path for every speaker.

Turn-taking clarity

While the engine accommodates natural dialogue, clear turn-taking without overlapping speech ensures the highest speaker diarisation accuracy.

Acoustic environment

Minimize room echo, air conditioning hum, and side conversations to maintain precise voice identification and transcription quality.