To transcribe multiple videos in Bonframe, prepare your local speech-recognition setup, open Auto ASR jobs, and drop a folder containing supported videos. Each file becomes a separate job. Review the resulting transcripts before using them for research, editing or translation.
Download Bonframe · Batch transcription feature
1. Prepare a representative folder
Keep a copy of your source material. Start with a few clear videos rather than a whole archive so you can check language selection and output quality on your hardware. The scanner currently recognizes MP4, MOV, MKV, WebM, M4V, AVI, MPEG and MPG video extensions; it is not an arbitrary audio-file folder importer.
Complete the runtime and ASR-model preparation offered by the installed app. The initial model downloads can require internet. Do not infer processing time from the number of files alone.
2. Choose the source-language setting
Check the recognition-language setting before submission. Correct recognition starts with intelligible audio and an appropriate language configuration. Translation into another language is a later task, not what ASR by itself does.
3. Drop a folder or inspect a list first
Open Auto ASR jobs. Drop one local folder at a time into this workspace; the app scans and immediately submits supported files. Do not assume there is a confirmation step after dropping.
For a preview before submission, use the folder picker, inspect the scanned list, then enqueue. If the app reports that the scan was truncated, divide the collection into smaller folders. The current UI scan request is bounded to 200 files and 8 directory levels.
4. Follow each job’s stages
Jobs proceed through import, audio extraction, ASR, text indexing and frame indexing. The queue processes work according to available local capacity, so multiple submissions are not a promise of parallel GPU execution.
Use the available pause, resume and individual task controls as needed. Inspect Needs attention jobs separately. A transcript appearing before frame indexing finishes does not mean the entire indexing job is complete.
5. Review projects, then search
Open completed projects and compare text with the source speech. Check names, dates, numbers, technical terms and overlapping speakers. Search the indexed material to locate relevant moments and continue editing from those results.
Transcription accuracy depends on the source audio and selected model; no fixed accuracy or speed is promised here. Use any text or subtitle export options available in the project; this workflow does not promise a folder-wide TXT/SRT export button.
Can I start from YouTube videos?
For permitted sources, use Media download, save the video artifacts into a folder, then run this workflow. Downloading, ASR and batch translation remain separate tasks with separate completion checks.