Obsidian Transcription Plugin: Voice Notes, Meetings, YouTube
On this page
SystemSculpt records audio and transcribes it to Markdown or SRT using hosted credits. Use your own AI free; Pro is $9/month for image and video generation, smart search and Studio.
What you need
- SystemSculpt AI, installed free from Obsidian's community plugins, on desktop or mobile.
- A SystemSculpt account with credits, set up under Settings → SystemSculpt AI → Account. Transcription uses SystemSculpt's hosted AI even if you chat with your own Claude Code or Codex.
- Your microphone and transcript preferences under Settings → SystemSculpt AI → Workflow.
Transcribe voice notes
- Run Toggle audio recorder from the command palette to open the recorder and start talking.
- Stop when you're done and wait for Recording saved before you close or switch away from Obsidian.
- With Transcribe automatically on in Workflow settings, transcription starts as soon as the recording ends. Turn on Keep source audio to keep the recording in your vault after transcription, and Insert transcript at origin to add the text back to the note or chat you started from.
Keep Obsidian open and in front while you record; recording in the background isn't guaranteed, on desktop or mobile.
Transcribe meetings and get a brief
For a meeting, record it as above or use an existing recording, then let Audio Processor write the brief:
- Run Open audio processor and choose Audio.
- Pick the recording from your vault, or upload one from your device.
- Choose Meeting brief, check the quote and start.
- Open the result: a short note with the outcome, summary, decisions, action items and open questions found in the meeting, linked to a full timestamped transcript.
Processing continues on SystemSculpt's service after the upload, so you can close Obsidian. Stop watching only closes the progress view; it doesn't cancel the job. Finished results wait for 7 days, so open Obsidian within that time to save them to your vault.
Read the transcript before you share the brief. Check names, dates, numbers and who agreed to what; a summary can hide whether a decision was made or only proposed.
Get a YouTube transcript into Obsidian
- Run Process YouTube video.
- Paste the full video URL.
- Choose Detailed note, Meeting brief or Clean transcript, and start.
SystemSculpt fetches the video's transcript on its service and saves a note under SystemSculpt/Audio Notes. A Detailed note has a summary, key points, decisions, action items, open questions and a link to the timestamped transcript, and its citations link to the matching moment in the video. Clean transcript gives you readable paragraphs without timestamps.
Transcribe an audio file to Markdown or SRT
Run Transcribe an audio file and choose a file. SystemSculpt accepts MP3, WAV, M4A, MP4, OGG, WebM and FLAC. Choose Markdown for an editable note, or SRT for a subtitle file. To make SRT the default, go to Workflow → Transcript output → Default file format and choose Subtitle file (.srt). You can also choose Convert Audio to Markdown from an audio file's menu.
Standard transcription takes files up to 500 MiB and 8 hours long. Audio Processor takes uploads up to 1 GB and 8 hours. The recorder has its own limits; the audio transcription docs list them.
After the transcript
A transcript is a note like any other: link it, search it by meaning, find related notes with Similar Notes, or attach it in SystemSculpt chat with @ and ask for the decisions and open questions. With Pro, a Studio workflow can read the transcript note and turn it into a summary or an image. Keep the transcript as the source and write summaries in a separate note, so you can check a quote against what was said.
Privacy and cost
Audio you choose to transcribe is sent to SystemSculpt's service for processing; see the Privacy Policy. Leave out recordings that shouldn't leave your device, and keep the original audio when you may need to verify it.
Transcription and Audio Processor use credits, not unlimited minutes; Audio Processor shows a quote before paid work. Eligible new accounts can claim 500 one-time starter credits; pricing lists Pro features and credit packs.
Questions
Does SystemSculpt transcribe offline? No. Recording happens in Obsidian, but transcription runs on SystemSculpt's hosted service and uses credits.
Can I make subtitles? Yes. Choose SRT when you transcribe an audio file. Audio Processor makes notes with a timestamped transcript instead.
Does it work on iPhone and Android? Yes, in Obsidian mobile. Keep Obsidian open while you record, and wait for Recording saved.
How long can a recording be? Standard transcription handles files up to 500 MiB and 8 hours, and Audio Processor up to 1 GB and 8 hours. The limits table has the details for each path.
Install SystemSculpt free, read the audio transcription docs, or see everything SystemSculpt does.