Uploading voice files and downloading transcripts

Skimle converts audio and video recordings into formatted text documents with automatic speaker identification, then analyses them alongside the rest of your project. Upload your interview recordings, focus groups, or other recordings as an Audio or video data source and Skimle transcribes them for you. If you keep the recording, it plays above its transcript and every quote can be played back from the moment it was said.

Supported formats

You can upload audio and video files in the following formats:

• Audio: AAC, FLAC, M4A, MP3, MPEG, OGG, WAV, WebM
• Video: AVI, MKV, MOV, MP4, WebM

Files can be up to 1 GB each. You do not need to extract the audio from a video first: a Zoom or Teams recording can be uploaded as it is.

Adding an audio or video data source

1. In your project's Data sources section, click New data source and choose Audio or video.
2. Drag and drop your recordings into the upload area, or click to browse your files.
3. Your files appear as pending.

At this stage, no credits are used. You can review your uploads before committing.

Choosing the data source options

The card below the upload area has four options. Each applies to this data source only, and to files confirmed from then on.

• Review transcripts: holds each transcript in a Review step so you can correct it before analysis. See Reviewing transcripts.
• Keep the recording: stores the audio or video beside its transcript so you can play it back. On by default. See Keeping or deleting recordings.
• Anonymise documents: pseudonymises the transcripts before analysis. See anonymising documents. A data source that anonymises never keeps recordings.
• Automatically analyse: sends each transcript to analysis as soon as it is ready.

Confirming transcription

Each pending file shows its duration and the credit cost for transcription. One minute of recording uses one credit; see pricing for how credits work.

1. Select the files you want to transcribe.
2. Review the total credit cost.
3. Click Confirm files to start transcription.

If you do not have enough credits, you will see a warning before confirming. Confirmed files are transcribed in the background, so you can close the page and come back later. You are notified by email when transcription is complete.

The transcript

Skimle produces a formatted transcript for each recording, which becomes a document in your project. It includes:

• Speaker labels: each speaker is automatically identified (Speaker 1, Speaker 2, etc.).
• Timestamps: each block of speech is marked with its start and end time.
• Formatted text: consecutive utterances by the same speaker are grouped into paragraphs for readability.

For example:

Speaker 1 [00:00:12 - 00:01:03]: I think the main challenge we faced was...

Speaker 2 [00:01:05 - 00:01:45]: That's interesting. Could you elaborate on...

The spoken language is detected automatically.

Playing a recording with its transcript

When a recording is kept, opening its document in the document view shows a player docked above the transcript. The player stays in place while you scroll and offers:

• Play and pause, with a scrubber that carries a tick for every quote coded from the recording
• Playback speed, mute and download
• Hide video, which folds away the picture of a video recording and leaves the playback bar

As the recording plays, the transcript follows along and underlines the word being spoken. Click any word in the transcript to jump the recording to that moment. The recording never starts playing on its own.

Playing a quote from its moment

Every quote that comes from a kept recording has a Play button. You find it on quotes in a document's insight panel, in the categories view and in the project overview.

Pressing Play opens the quote's document if you are not already on it, moves the player to a second before the quote begins and starts playback, so you can hear exactly how the words were said. Use it to check a quote's wording before you use it verbatim, or to hear tone and emphasis that a transcript cannot show.

Reviewing transcripts

Turn on Review transcripts on the data source if you want to check and correct the automatic transcripts before they are analysed. Confirmed recordings then wait in a Review step where you can fix speaker labels or wording; once you approve them, they continue to analysis. With review off, transcripts flow straight through.

Keeping or deleting recordings

With Keep the recording on, the audio or video is stored privately in Skimle's EU infrastructure, and only members of the project can play or download it. This is what makes playback and quote playback possible, and it takes up storage space.

With the option off, the recording is deleted by the time its transcript is ready, and nothing can be played back afterwards. Choose this when your ethics approval or data management plan requires recordings to be destroyed after transcription.

A recording is never kept when:

• the data source anonymises its documents, because a voice can identify a person however well the text is pseudonymised
• your team administrator, or your organisation's umbrella account, has disabled keeping recordings; the option is then not offered

You can delete a kept recording at any time: from the document's panel, from the data source's processed files, or for every file in the data source at once. Deleting a recording keeps its transcript and analysis.

Analysis and privacy

Transcripts are analysed in place, alongside every other document in the project, so there is no separate download step. You can still export the finished analysis, including the transcripts, at any time.

All transcription runs on EU-hosted infrastructure, and your recordings and transcripts are not used to train AI models.

Once your recordings are transcribed, add an analysis to code them into themes and categories.