Setup & provisioning
Namespace identity, verified hardware, installed model weights and the processing profile used for imports in this session.
Active Namespace
Namespace ID: —
Root Path: —
Storage Status:
—
Storage usage unavailable.
Workspace & Revision Storage Accounting
Storage accounting scan not yet performed.
Hardware Detection
CPU Cores: Detecting...
System RAM: Detecting...
FFmpeg Engine: Detecting...
Local Model Weights Audit
Auditing local model weights...
Model Cache & Search Diagnostics
The following cache and workspace locations are inspected for local weights:
Other model weights discovered in system cache:
- Auditing local model weights and cache repositories...
Model Installation Guide
SPA maintains a strict zero implicit downloads policy during transcription. Download model weights ahead of time:
uvx --from=huggingface_hub hf download Systran/faster-whisper-large-v3
huggingface-cli download Systran/faster-whisper-large-v3python3 -c "from faster_whisper import download_model; download_model('large-v3')"
Place model files (model.bin, config.json, tokenizer.json, vocabulary.json) in:
<namespace_root>/models/whisper-large-v3/
Select Processing Profile
Checking active profile...
The selected profile applies to recordings you import in this session. It is not saved to the workspace.
Recordings Library
Add recordings
Drop files or choose them from disk. Each supported file is uploaded in order and gets its own result.
📁 Drag & Drop Audio/Video Files Here
Supports WAV, FLAC, Opus, OGG, MP3, M4A, AAC, MP4 and MKV. Every supported file is uploaded in order and gets its own result.
Keyboard: press Enter or Space to choose files.
Import Queue
Ready to upload.
Not loaded yet.
| Recording | Duration | Status | Stage & Progress | Reviews | Actions |
|---|
No recordings yet
Import audio or video to start a review. Every supported file becomes its own library entry with measured duration, stage and review facts.
Choose a review project
Projects are ordered by most recent edit. Viewing a project does not move it.
No review projects yet
Open a recording from the Library to start the first review.
Review Workspace
Annotation lane: click to mark a point, drag to mark a range. With the lane focused use ← → to move (Shift ×10), Space to start/finish a range, Enter to label the caret.
Timeline settings
Annotation filters
Bookmarks & Annotations
0 activeManage Categories
No bookmarks or annotations found.
Review transcript
Filters & display
Use the source audio to check each utterance. Press J/K to move through filtered events, or through the current search matches.
Matches text, speaker and note, combined with the filters above. J/K cycles matches. Search terms are included in the review link.
Audio processing
Preview an exact source interval, compare original with enhanced audio, and review changed candidates.
Audio processing Preview an exact source interval, compare both versions, then apply the same settings to the full recording.
Enhanced
Preview not createdA/B difference review Aligned synthetic guidance for the latest preview pair. Ordinal and uncalibrated: it is not perceptual quality, speech audibility, or a human-benefit result.
Create a preview in Audio processing to compare an exact source interval.
Cache provenance and truthful limits
- Cache key
- —
- Cache state
- —
- Source SHA-256
- —
- Matched SHA-256
- —
- Manifest SHA-256
- —
- Parameters
- —
- Coordinates
- —
- Engine
- —
- Alignment
- —
- Levels
- —
Changed-region candidates
No changed-region candidates are available. Create or refresh a preview pair first.
Exact-interval A/B
A is the original source slice; B is the matched slice at the same source-sample boundaries. Switching keeps both buffers running so the source position cannot drift.
No A/B pair prepared.
A · 0 ms / 0 ms
Reviewed notes
Saving a note creates an undoable review revision through the mutation envelope. The comparison artifact itself is read-only and never rewritten by a review.
Processing operations log Recent audio preview, enhancement, and restoration operations for this recording.
Revision history
Inspect and preview revision history DAG and lineage.
Revision history
Loading retained revisions…
No retained revisions yet.
Select a revision to preview it. Preview is read-only and never moves the project head.
Exports
Published revisions provide canonical links and direct downloads (Opus, MP3, FLAC, WAV, M4A, JSON, SRT). Video renders bind the published revision and audio track selected when queued.
Published revisions
No published revisions in this namespace yet. Publish a review from the Review tab to create a canonical transcript and audio download links.
Queued renders
No exports queued in this namespace. Attest & publish a review, then choose Export Synced Video to queue an MP4/WebM render. Published transcripts download from the publication receipt.
Durable export receipts
No durable export receipts in this namespace yet. Exporting video or audio, or downloading a project document, creates one.
Collections
A collection pins a noun: a project head, a project snapshot, an export receipt or a bookmark. Pinned snapshots keep their identity when the project is edited later; divergence is shown, never rewritten.
Select a collection to inspect its members, pinned snapshots and manifest exports.
—
Members
No members yet. Add a project, snapshot, export receipt or bookmark below.
Add a member
Create a bookmark
Bookmarks pin a project snapshot and an optional transcript position. Once created they can be added to any collection.
Manifest exports
No manifest export yet. Exporting writes an immutable, content-addressed JSON manifest of every resolved member.