Create, review and download transcripts in Feltiv

Extracting text from speech using Transcription feature

Transcription turns the speech in your audio or video files into accurate, time-coded text. This article takes you through the whole journey: create the project, follow its jobs, review the transcript and download it.

Info
You provide: audio or video files, or your own transcript if you choose the Only Review workflow.
You get: an editable, time-coded transcript that you can download as JSON, TXT, PDF, DOCX, CSV, SRT or VTT.
Review: yes, in the review editor.

Not sure whether you need transcription, subtitling or captioning? Read the article that compares the three.

Create a transcription project

  1. Select New in the left sidebar. On the Start a project page, select the Transcription card. Feltiv shows what the service does, the languages it supports and the file formats you can upload. Select Start Project.
[[IMAGE 1 - kb-screenshots/captioning/00-start-a-project-wide.jpg: the Start a project page. It also shows the Audio Description card, which is not ready to publish, so crop it out or retake the screenshot. Delete this line after you add the image.]]
  1. In Start a Transcription Project, choose a workflow. Transcription with AI generates transcripts for your content from scratch using AI. Only Review skips AI generation, so you bring your own transcripts and use Feltiv for quality checking. If you already have a transcript, read Already have a transcript? Use Only Review below.
  2. Enter a Project Name. It must be 3 to 100 characters and cannot contain single or double quotes. You can add a short Description. Select Start.
[[IMAGE 2 - kb-screenshots/transcription/01-start-transcription-window.jpg: the Start a Transcription Project window. Delete this line after you add the image.]]
  1. On the create page, choose the Source Language, which is the language spoken in your content, and a Model. Both are required. You can search the language list. The models you can choose depend on the language, and each one shows its price in credits, so you can see the cost before you create the project.
  2. Add your audio or video files. Choose Upload (from your device), OneDrive, Google Drive, Amazon S3 or Azure BLOB. Each file must be under 2 GB, and you can upload up to 20 files at a time. The upload panel lists the supported file types, which are WAV, MP4, MPEG, MP3, MOV and MPG for transcription. See the article on uploading files for the upload messages.
  3. Select Create Project. Feltiv takes you to the project's details page.
[[IMAGE 3 - kb-screenshots/transcription/03-create-page-model-selected.jpg: the transcription create page. Delete this line after you add the image.]]

Already have a transcript? Use Only Review

Choose Only Review when you already have a transcript and want it reviewed and quality-checked in Feltiv. You can have the review done by your own team or assign it to a professional QC vendor through the Feltiv QC Vendor Marketplace.
  1. Choose the Source Language. The card shows the cost for the Without AI model.
  2. Upload your media files first.
  3. Each media file gets one slot for each source language. Add your transcript in SRT, XLSX or CSV format. Use Download sample XLSX file or Download sample CSV file to see the layout.
  4. Every media file needs its transcript before the form can be submitted. Select Create Project when they are all in.

Track your jobs

Open the project to see one row for each file. Use Group by Content to nest a transcript's translation jobs under its media file, or Group by Job for a flat list. Each row shows the model used, the credits debited for the job, a progress tag and a review status. Statuses, the progress tag and Retry work as described in the article on the project details page.

Review and complete

  1. When the review status shows that the AI output is ready, select Review. For an Only Review job, the editor opens your uploaded transcript.
  2. Edit lines, speakers and timecodes with the tools in the review editor. The Speakers button shows every speaker in one place and lets you merge duplicates.
  3. If someone reviewed the file outside Feltiv, you can upload the reviewed SRT or VTT file back into the job, preview the changes, and confirm it as the Final version. See the article on the review editor.
  4. Select Mark as Complete and choose Language QC, Timecode QC or both.
To translate a finished transcript, select Add Translation on the job, or + Translation in the project header. Both need permission.

Download your transcript

Select the download icon on the job and choose a format. Under Select Type, pick Paragraph or Phrase/Sentence to narrow the list of formats. See the article on downloading your output for the CSV, offset and cloud storage options.

Need help?

If a job fails again after a retry, write to support@feltiv.com with the project name and the file name.