The Problem with Meeting Transcripts
Standard transcription tools give you a wall of text with no indication of who said what. For business meetings, this is almost useless. You need to know:
- -Who made which decision?
- -Who committed to which action items?
- -Who raised which concerns?
The Solution: Speaker Diarization
Speaker diarization is AI technology that automatically identifies different speakers in an audio recording. Combined with transcription, you get transcripts that look like this:
[Sarah - CEO]: Let's review the Q4 targets.
[Mike - Sales]: We're currently at 85% of target.
[Lisa - Marketing]: The new campaign should help close the gap.
[Sarah - CEO]: Great. Mike, can you send the updated forecast by Friday?
Step-by-Step Guide
Record Your Meeting
Use Zoom, Teams, or any recording software. For best results, ensure each participant has a good microphone and minimize background noise.
Upload to SplitBySpeakers
Drag and drop your audio or video file. We support MP3, WAV, M4A, MP4, and more. Files up to 100MB are supported.
AI Identifies Speakers
Our AI analyzes the audio and automatically identifies up to 5 different speakers, creating separate audio tracks for each.
Name Your Speakers
Replace "Speaker A" and "Speaker B" with actual names. Listen to a sample from each track to identify who's who.
Download & Transcribe
Download the separated tracks and use your preferred transcription service (like Whisper or Otter.ai) on each track for speaker-labeled transcripts.
Business Use Cases
Meeting Minutes
Create accurate records of who said what for compliance and documentation.
Team Collaboration
Share searchable transcripts so team members can find their action items.
Action Item Tracking
Clearly attribute commitments to specific team members.
Client Calls
Document client requests and your team's responses accurately.
Pro Tips for Better Results
- Use quality microphones: Individual headsets work better than a shared conference room mic.
- Minimize crosstalk: When possible, avoid multiple people speaking at once.
- Record in high quality: Use WAV format if possible, or high-bitrate MP3 (at least 128kbps).
- Identify speakers early: Have each person introduce themselves at the start for easier identification.