Datasets
| Dataset | Tasks |
|---|---|
| Dataset | Tasks |
|---|---|
MultiMedia-TerminalBench (MMTB): a benchmark of 105 realistic multimedia-file tasks in persistent terminal workspaces, across 5 meta-categories grounded in paid practitioner workflows.
harbor run -d mmtb/multimedia-terminalbench| Task |
|---|
| Task |
|---|
mmtb/warehouse-sku-pack-audit |
mmtb/external-mic-sync-repair |
mmtb/semantic-image-retrieval |
mmtb/fugal-subject-entry-labeling |
mmtb/cursor-deictic-thumbnails |
mmtb/audience-ringtone-detection |
mmtb/crm-compliance-audit |
mmtb/spoken-vs-displayed-claim |
mmtb/polyphonic-piano-feedback |
mmtb/semantic-chaptering |
mmtb/ornament-classification-detection |
mmtb/emotional-arc-match |
mmtb/adr-edit-detection |
mmtb/stereo-channel-flip-repair |
mmtb/robotics-demo-command-audit |
mmtb/dialogue-exchange-match |
mmtb/receipt-photo-to-json |
mmtb/boss-cooldown-cheat-audit |
mmtb/multi-mic-bleed-attribution |
mmtb/travel-clip-retrieval |
mmtb/musical-mood-shot-pick |
mmtb/broadcast-package-edit |
mmtb/narration-visual-align |
mmtb/safe-single-cue-keep |
mmtb/narration-drift-qc |
mmtb/birthday-money-shot |
mmtb/slack-action-extraction |
mmtb/blind-audition-match |
mmtb/phone-level-pronunciation-errors |
mmtb/b-roll-pool-assignment |
mmtb/screenshare-deictic-grounding |
mmtb/caption-nonspeech-enrichment |
mmtb/narration-mars-rover |
mmtb/accessibility-sync-audit |
mmtb/long-form-clip-miner |
mmtb/design-review-version-approval |
mmtb/tempo-drift-detection |
mmtb/interview-music-ducking-audit |
mmtb/coop-voice-callout-audit |
mmtb/audio-visual-dub-detection |
mmtb/take-tone-reaction-pick |
mmtb/av-desync-offset-repair |
mmtb/lecturer-visual-term-ref |
mmtb/lexical-stress-classification |
mmtb/articulation-deviation-detection |
mmtb/animation-narration-audit |
mmtb/interview-srt-refine |
mmtb/sports-broadcast-events |
mmtb/caption-speech-mismatch |
mmtb/blood-test-pdfs-to-csv |
mmtb/dead-air-removal |
mmtb/comping-chord-substitution |
mmtb/phoneme-confusion-patterns |
mmtb/page-photo-to-text |
mmtb/delivery-clip-defect-triage |
mmtb/speaker-roster-identification |
mmtb/chapter-repair |
mmtb/creator-voiceover-lipsync-mismatch |
mmtb/batch-media-qc-audit |
mmtb/traffic-cam-incident-audit |
mmtb/signal-based-qc-report |
mmtb/cross-channel-privacy-leak |
mmtb/tutorial-edit-recreation |
mmtb/near-duplicate-frame-dedup |
mmtb/spoken-decision-cell-ref |
mmtb/av-desync-detection |
mmtb/game-outcome-qa |
mmtb/piano-practice-feedback |
mmtb/violin-intonation-detection |
mmtb/invoice-estimate-pdfs-to-xlsx |
mmtb/lecture-demo-clip-extract |
mmtb/polyrhythm-accuracy-detection |
mmtb/prosody-multi-dim-selection |
mmtb/multicam-active-speaker-cut |
mmtb/code-review-comment-attribution |
mmtb/stream-alert-ack-audit |
mmtb/constant-offset-srt |
mmtb/speaker-action-attribution |
mmtb/question-statement-intonation |
mmtb/prosody-take-selection |
mmtb/string-quartet-mistake-attribution |
mmtb/constant-hum-attenuation |
mmtb/bug-repro-claim-audit |
mmtb/speedrun-input-tamper-detect |
mmtb/quote-clip-retrieval |
mmtb/vfr-drift-repair |
mmtb/game-alert-mismatch |
mmtb/call-center-disclosure-audit |
mmtb/pronunciation-error-flagging |
mmtb/lipsync-drift-correction |
mmtb/av-identity-leak-detect |
mmtb/cooking-instruction-alignment |
mmtb/multi-utterance-pronunciation-errors |
mmtb/partial-srt-resync |
mmtb/proof-step-note |
mmtb/mock-call-automation |
mmtb/dub-speaker-mismatch |
mmtb/design-review-approval-audit |
mmtb/narration-music-ducking |
mmtb/podcast-episode-assembly |
Displaying 100 of 105 tasks