How to correct speaker names and labels
Rename speakers and check them against the audio.
Speaker labels make a conversation easier to follow, but the names should still match the people who actually spoke.
These tools are available in HoldSay.
Generic labels become people: rename a speaker once and the whole transcript follows.
Rename a generic speaker
After a recording is transcribed, HoldSay may identify people with labels such as Speaker 1 and Speaker 2. Your goal is to replace those generic labels with names or roles that make the conversation readable to you and to anyone who receives the note. Start by opening the recording in HoldSay and choosing Speakers. This view brings the speaker labels and their passages together, so you can correct the identity without hunting through the full transcript first.
Generic speaker labels can be renamed directly in the completed note.
- Choose the generic label you want to correct, such as
Speaker 1. - Enter the person's name, or enter a clear role such as
Interviewer. - Save the change, then scan the other passages assigned to that label.
Use a person's name when you expect to search for that person again. For example, a project discussion is easier to review later when passages are labeled with the participant's name. Use a role when the identity is not important to the note, such as when you only need to distinguish an interviewer from a guest. After saving, the label should appear consistently on the passages assigned to that speaker in the note. Read several passages, especially near the beginning and end of the recording, to confirm that the label represents one person rather than a sequence of different voices.
Connect a recurring person
If you record the same person regularly, connect or save their name to a speaker profile instead of treating each recording as a separate identification task. A speaker profile can carry the same identity into future recordings in your local library. This is useful for recurring meetings, interviews, or conversations where one person appears in more than one note.
In the recording's Speakers view, work from a label you have already checked against the audio. Save or connect that name to a speaker profile when you are confident that the label belongs to the same person throughout the recording. The practical result is that the person has an identity HoldSay can use again in later recordings held in your local library.
Connect a recurring voice to one saved local speaker profile.
When the current recording does not provide a reliable enough example, use Teach HoldSay who is speaking to create the profile from a clearer sample. Choose a sample where the person speaks distinctly enough to separate their voice from others. A short acknowledgement, an interruption, or an overlapping response is a poor basis for deciding an identity because it may be hard to tell who is speaking.
- Confirm the label with several passages before connecting it to a profile.
- Use a clearer sample when the voice is brief, interrupted, or mixed with another speaker.
- Check future recordings when the profile appears, particularly where speakers sound similar.
A profile reduces repeated naming work, but it does not remove the need to review important passages. If a later recording contains a fast exchange or unclear audio, replay the relevant boundary before relying on the label.
Check difficult speaker handoffs
Most speaker label errors happen at a handoff, the point where one person stops and another starts. These boundaries matter most when you plan to quote a passage, share a transcript, or use the note to recall a decision. In Speakers, click the timecode beside the passage you want to check. Playback jumps to the matching moment in the recording, allowing you to compare the label with the audio rather than guessing from the words alone.
- Click the timecode beside an uncertain passage in Speakers.
- Listen to the speech immediately before the labeled passage.
- Continue through the boundary and listen to the beginning of the next speaker's words.
- Decide whether the label belongs to the voice you hear, then correct it when needed.
Review the boundary where one person stops and the next person begins.
Do not assume that a paragraph belongs to one speaker just because most of it does. Short acknowledgements such as a brief response can belong to a different person than the surrounding paragraph. Interruptions can also move the boundary earlier or later than expected. Listening on both sides of the timecode makes these changes easier to hear.
Prioritize passages that will be quoted or shared, rather than trying to treat every uncertain moment as equally important. Overlapping speech may still require a human decision. When two voices are audible at once, use the surrounding conversation and the audio to decide which person should receive the passage label. If the recording does not make that clear, avoid presenting the label as more certain than the audio supports.
Review names before sharing
Before sharing a transcript, review the names as well as the words. A correctly transcribed sentence can still mislead a reader if it is attributed to the wrong person or if a name is spelled differently in the speaker label and the transcript text. Start by searching each important name and reading a sample of the passages where it appears. This shows whether the same identity is used consistently across the note.
Focus your review on places where two people speak quickly, interrupt each other, or return to a topic after a pause. These are the places where a generic label may have been assigned across a speaker change. Open the relevant passage in Speakers, use its timecode to replay the audio, and compare what you hear with the displayed name. The result you want is simple: the label, the transcript text, and the voice in the recording should agree.
Named speakers use real photos and remain easy to distinguish in later review.
- Search every important person's name and read a representative set of passages.
- Check quick exchanges and interruptions against playback.
- Correct name spelling in both speaker labels and transcript text.
- Keep the audio with the note when a future reader may need to verify a quotation.
Completed notes in History can show views such as Raw, Speakers, Polish, or Summary when that view applies to the source. For speaker review, return to Speakers when you need to inspect attribution, then use the recording playback to resolve uncertainty. This keeps the final note useful for reading while preserving a way to verify a quoted passage later.
Common questions
Can I rename Speaker 1 in HoldSay?
Yes, in HoldSay for Mac. Open Speakers, choose the generic label, and replace it with the person's name or a clear role.
Does changing a speaker name change every passage with that label?
The renamed label is applied consistently to the passages assigned to that speaker in the note. Scan those passages after saving to confirm they all belong to the same person.
How do I check an uncertain speaker label?
In the Speakers view, click the timecode beside that passage. Listen before and after the boundary, then decide who spoke based on the audio.
If a label on this page does not match the app, email support@holdsay.com. Include Mac or Windows and the HoldSay version shown in Settings.