n8n Workflow: Batch AI Voice-Overs with Gemini + Telegram

Turn queued topics into batch-ready AI voice-overs—then get human approval in Telegram

This n8n workflow automatically pulls queued narration topics from your n8n Data Table, generates a structured script with Google Gemini, produces WAV voice-overs via Gemini TTS (with backups), optionally runs a listen-back QA check, and sends the audio + script files to Telegram for approve/redo decisions.

What this workflow does

• Scheduled batch run: Executes every morning, ensures the vs_queue Data Table exists, and pulls the next queued topics up to your configured batch size.
• Script generation (Gemini + JSON): For each topic, it builds a niche- and format-specific prompt and uses Google Gemini to generate a structured JSON script, with a fallback Gemini model if needed.
• Script validation & improvement: It checks the script for length, hook, scene structure, clichés, visuals, and facts. If enabled, it can ask Gemini to fix issues once and keep the better version.
• Voice-over + captions: Generates narration as a WAV using Google Gemini TTS (backup TTS supported). It derives scene timing from real pauses and builds SRT captions.
• Optional listen-back QA: Transcribes the audio with Gemini and compares the transcription to the script to compute a word-level match score and identify missing/extra words.
• Write results back + start review: Stores checks, listen-back score, package JSON, and metadata in the vs_queue row, then triggers a separate review run for each voice-over.
• Telegram approval loop: Sends the audio and a detailed script text file to Telegram and waits for approval or redo, then marks the row as approved when accepted.

Use cases

• Automating weekly voice-over production for SaaS marketing content batches.
• Scaling “write → narrate → QA → human approve” pipelines without manual file handling.
• Adding quality control via listen-back scoring before publishing audio scripts.

Technical details

• n8n nodes/tools: if, set, code, telegram, data table, sticky note
• Integrations: Google Gemini for script generation + transcription, Gemini TTS for WAV narration (with backups), Telegram for approve/redo workflow