Are Speaker A and Speaker B real identities?

No. They are automatic labels for voice groups, not verified personal identities.

Can I rename a speaker in this interface?

Not yet. Download labeled TXT and edit the names in a text editor after checking the source audio.

Can I merge two labels?

There is no merge control. If listening confirms that two labels belong to the same person, correct them in your downloaded TXT.

What if two people share one label?

Listen to the affected passage, split the turns in your document and assign each turn manually. Avoid global replacement until mixed turns are corrected.

Will it separate people who speak at the same time?

Overlapping voices may be missed or assigned incorrectly. Review interruptions and short responses against the recording.

What files and limits apply?

Upload MP3, M4A, WAV or MP4 up to 4 MB. Only the first 10 minutes are processed, and sign-in is required.

Can I export without speaker labels?

Yes. Turn off Show speaker labels, then copy or download the plain TXT. Timed SRT is available when the provider supplies utterances.

Is there a separate speaker-label charge?

This version does not deduct ToTextly credits for file transcription or its automatic speaker labels. A recurring free allowance has not been established.

How do I delete the provider transcript?

Use Delete provider transcript in the result controls. Deletion is subject to AssemblyAI retention and backup policies. Download the result first if you need to keep a copy.

Transcription with Speaker Labels

Separate voices in an uploaded recording with automatic speaker diarization. Review speaker-labeled text and export TXT or SRT. Sign-in required.

Speaker labels distinguish voices; they do not identify real people and can be wrong. Download your result before leaving this page. ToTextly does not provide a saved file-transcript library.

What speaker diarization actually labels

Speaker diarization groups speech by voice and places a label such as Speaker A or Speaker B on each turn. These labels distinguish voices; they do not reveal or verify a person’s name. A label can be wrong, so check the audio before attributing a quotation.

This tool transcribes an uploaded recording with automatic speaker labels enabled. It is useful for interviews and conversations with clear turns. It does not identify faces in a video, match a voice to a contact, or establish who attended a meeting.

How to get a speaker-labeled transcript

Sign in, choose an MP3, M4A, WAV or MP4 file up to 4 MB and confirm permission to process it. Select Transcribe file and wait for recognition to finish. Only the first 10 minutes are processed.

The result shows a labeled version by default. Switch Show speaker labels off to read the plain transcript. Copy or Download TXT uses the currently selected view. Download SRT is available when the provider returns timed utterances.

Rename speakers after checking the recording

The current interface has no rename control. To replace Speaker A with a real name, download the labeled TXT, listen to several turns and confirm the identity before using find-and-replace in a text editor. Do not guess a name from a voice alone.

Keep the original labeled copy as a reference. If several people share one automatic label, replacing that label globally will introduce incorrect attributions. Fix those mixed turns individually before applying a name to the remaining passages.

Merge split labels and repair mixed speakers

One person may receive two labels after a long pause, a microphone change or a shift in delivery. Compare the corresponding passages with the recording. If they belong to the same speaker, replace the second label in your downloaded document. There is no automatic merge button in this version.

The opposite error merges two voices under one label. Locate the change of voice in the source, split the affected text and assign the correct label manually. Short acknowledgments and interruptions need particular care because they provide little voice context.

Where speaker labels work best

Distinct voices taking turns with clear microphones are easier to review than crowded conversations. A quiet interview with full answers is a better starting point than a fast debate with frequent interruptions. Labels remain estimates even in a clean recording.

Overlapping speech can cause words to disappear or move to the wrong person. Music and room echo can obscure differences between voices. If attribution matters for publication, check each quoted turn and mark uncertain passages rather than treating generated labels as evidence.

Exports, privacy and current limits

Labeled TXT separates turns for editing interview questions, research notes or a podcast outline. Plain TXT removes the visible labels. SRT includes speaker text and utterance timing when available; long turns may need splitting in a subtitle editor. No VTT export is currently provided.

Files are sent to AssemblyAI for transcription. You can request deletion of the provider transcript from the result controls, subject to provider retention and backup policies. This version has no saved transcript library, no translation feature and no live microphone capture. File transcription currently does not deduct ToTextly credits.

Frequently asked questions

Are Speaker A and Speaker B real identities?: No. They are automatic labels for voice groups, not verified personal identities.
Can I rename a speaker in this interface?: Not yet. Download labeled TXT and edit the names in a text editor after checking the source audio.
Can I merge two labels?: There is no merge control. If listening confirms that two labels belong to the same person, correct them in your downloaded TXT.
What if two people share one label?: Listen to the affected passage, split the turns in your document and assign each turn manually. Avoid global replacement until mixed turns are corrected.
Will it separate people who speak at the same time?: Overlapping voices may be missed or assigned incorrectly. Review interruptions and short responses against the recording.
What files and limits apply?: Upload MP3, M4A, WAV or MP4 up to 4 MB. Only the first 10 minutes are processed, and sign-in is required.
Can I export without speaker labels?: Yes. Turn off Show speaker labels, then copy or download the plain TXT. Timed SRT is available when the provider supplies utterances.
Is there a separate speaker-label charge?: This version does not deduct ToTextly credits for file transcription or its automatic speaker labels. A recurring free allowance has not been established.
How do I delete the provider transcript?: Use Delete provider transcript in the result controls. Deletion is subject to AssemblyAI retention and backup policies. Download the result first if you need to keep a copy.