Workflow Recipes

Workflow Recipes are pre-architected DAG topologies tailored for high-frequency productivity scenarios. Leveraging DuRT 3.0’s Directed Acyclic Graph engine, you can build single-pass pipelines, parallel multi-branch execution, and multi-destination delivery.

---

Recipe 1: Quick Media Transcription → Dual Subtitle Export

Use Case: You have a local interview, podcast, or meeting recording and want to generate accurate subtitles with word-level timestamps in both SRT and JSON formats.

DAG Topology

[ mediaInput (File / Link Ingest) ]
               │
               ▼
[ mediaASR (Transcription & Diarization) ]
               │
               ▼
[ exportArtifact (Export SRT + JSON) ]

Key Configurations

• mediaInput: Select local media files (MP4, MOV, MP3, WAV, etc.) or paste streaming URLs;
• mediaASR:
  • Word-level Timestamps: ✅ Enabled (millisecond precision);
  • Speaker Diarization: ✅ Enabled (segments Speaker 1 / Speaker 2);
• exportArtifact: Enable both SRT (for video player playback) and JSON (for editing scripts and metadata).

---

Recipe 2: Parallel Meeting Pipeline: Refined Subtitles + AI Minutes + Notion Sync (DAG Branching)

Use Case: You recorded a 1-hour strategic meeting and want both polished, filler-free subtitles and an AI-generated executive summary with action items synced straight to your team’s Notion database.

DAG Topology

                   [ mediaInput (Import Meeting Audio) ]
                                    │
                                    ▼
                   [ mediaASR (ASR & Speaker Diarization) ]
                                    │
               ┌────────────────────┴────────────────────┐
               ▼                                         ▼
    [ subtitleRefinement ]                     [ contentCreation ]
 (Filler Removal & Punctuation)             (Summary & Action Items)
               │                                         │
               ▼                                         ▼
    [ exportArtifact (Local SRT) ]           [ exportArtifact (Notion Sync) ]

Key Highlights & Configurations

• Zero Idle Waiting: The timed subtitle track (subtitleTrack) emitted by mediaASR feeds concurrently into both formatting and summarization nodes, cutting total runtime by over 40%;
• subtitleRefinement: Enable filler-word removal ("um", "like") and set maximum subtitle line length to 32 characters;
• contentCreation: Select the [Meeting Minutes] preset to extract key decisions, risks, and assigned action owners;
• exportArtifact: Connect to your Notion database to automatically append structured database cards.

---

Recipe 3: Automated Video Localization & AI Voice Dubbing

Use Case: You have an educational video and want to generate translated bilingual subtitles alongside a natural, AI-dubbed voiceover audio track.

DAG Topology

[ mediaInput (Import Video) ] ➔ [ mediaASR (Accurate ASR) ] ➔ [ fullTranslation (Multilingual) ]
                                                                             │
                                           ┌─────────────────────────────────┴─────────────────────────────────┐
                                           ▼                                                                   ▼
                                [ textToSpeech (AI Voiceover) ]                                   [ exportArtifact (Bilingual SRT) ]
                                           │
                                           ▼
                                [ exportArtifact (Dubbed MP3) ]

Key Configurations

• fullTranslation: Set target language to English (or any of 50+ supported languages) with timestamp alignment enabled;
• textToSpeech: Choose a high-fidelity neural voice from OpenAI TTS or OneTTS;
• Dual Deliverables: Instantly receive both the time-aligned bilingual .srt file and the synthesized .mp3 voiceover track, ready to drop into Final Cut Pro or Premiere.

---

Recipe 4: Live Meeting & Lecture Bilingual Floating Subtitles

Use Case: Attending a foreign-language Zoom conference or watching an untranslated YouTube lecture, requiring real-time on-screen bilingual subtitles.

DAG Topology

[ monitorCapture (System Audio Capture) ] ➔ [ realtimeASR (Streaming ASR) ] ➔ [ realtimeTranslation (Live Interpretation) ]

Key Configurations

• monitorCapture: Toggle “System Internal Sound” and activate hardware/DSP Acoustic Echo Cancellation (AEC);
• realtimeASR: Select Apple On-Device Speech for zero data usage, or OneASR streaming gateway;
• realtimeTranslation: Select macOS 15 on-device neural translation or DeepSeek API, toggling the floating window into bilingual side-by-side mode.

---

Recipe 5: Enterprise Multi-Cloud Archival Pipeline

Use Case: Legal, consulting, or corporate teams requiring all client interview transcripts and raw recordings to be archived in S3-compatible cloud storage, mirrored to team WebDAV drives, and logged via Webhook.

DAG Topology

[ mediaInput ] ➔ [ mediaASR ] ➔ [ contentCreation ]
                                       │
                 ┌─────────────────────┼─────────────────────┐
                 ▼                     ▼                     ▼
       [ S3 Object Storage ]    [ WebDAV Drive Sync ]   [ Custom HTTP Webhook ]

Key Configurations

• exportArtifact #1: Select your S3 account and set destination bucket to meeting-archives/2026/;
• exportArtifact #2: Select WebDAV and synchronize to your Synology NAS or Nextcloud shared folder;
• exportArtifact #3: Connect to your company’s internal webhook or n8n endpoint with standardized JSON payloads.