n8n File Conversion Templates: 5 Ready-to-Import Flows

TL;DR - An n8n workflow template is a complete automation blueprint exported as JSON — import it, add credentials, and the workflow runs without manual node configuration - These five templates cover the highest-failure conversion tasks in production: PDF→text, DOCX→PDF, image batch resize, audio transcode, and bulk multi-format - Every template calls a REST conversion API over standard HTTP — no proprietary nodes, no SDKs, adaptable to any multipart-upload service - The most common failure mode in file conversion workflows is binary data silently dropped between nodes; each template explicitly names the binary property to prevent this - n8n ships with more than 400 native integrations (n8n.io/integrations), but file conversion sits at the intersection of binary data, async APIs, and HTTP encoding — more edge cases than most nodes expose

You need a file converted inside an n8n automation. Sounds simple.

It isn't. The free API throttles at 50 requests per day. ffmpeg in Docker works until it doesn't, usually on the most complex file in your batch. The StackOverflow answers for "n8n multipart binary upload" reference a version from three years ago. You spend a Friday afternoon on infrastructure instead of the pipeline you actually needed.

These five n8n workflow templates solve that. Each is a tested, importable JSON file for a specific conversion task that breaks real pipelines: PDF text extraction, DOCX to PDF, image batch resize, audio transcoding, and bulk multi-format conversion. They're built for teams running document pipelines, media operations, or content automation — not hobbyist tutorials that work once with a clean file and an unconstrained API key.

Who this is NOT for: if you're converting a handful of files a week manually, a desktop tool is faster. These templates pay off when conversion sits inside a pipeline — triggered by a webhook, chained to AI processing, or scheduled against a backlog.

---

What Are n8n Workflow Templates and Why Import JSON?

An n8n workflow template is a pre-built automation blueprint saved as a JSON file, containing every node, connection, credential reference, and parameter needed to run the workflow. Import it into any n8n instance — cloud or self-hosted — and the complete workflow appears in your canvas, ready to configure.

The community template library at n8n.io/workflows holds thousands of shared workflows. Most are conceptual or outdated: they show a node structure without error handling, without correct binary data configuration, and without the async polling patterns that conversion APIs actually require. These five are tested against live APIs in 2026, with explicit error branches and correct Content-Type headers.

The real value of a working n8n workflow json example: you see exactly how Authorization headers are formatted for multipart uploads, how binary data flows from an HTTP Request node to the next step, and how to poll for job completion on APIs that return a job ID rather than an immediate result. Building that from scratch takes hours. Importing it takes ninety seconds.

n8n ships with more than 400 native integrations (n8n.io/integrations, 2026). File conversion isn't one of them — it lives in the gap between binary data handling and REST API calls, and that gap has more failure modes than most teams expect before they hit them.

---

The 5 Templates at a Glance

Pick by pain point, not complexity. Document pipelines: start with PDF or DOCX. Media operations: image or audio. Standardizing a mixed archive: bulk multi-format.

---

How to Import Any n8n Workflow JSON

Importing takes under two minutes and the process is identical on cloud and self-hosted n8n.

1. Copy the JSON from the download bundle below or any code block in this article
2. Open n8n → click + (Create New Workflow) or open an existing one
3. Click the three-dot menu (top right) → Import → From JSON
4. Paste the JSON → Import
5. Double-click any HTTP Request node → add your API key: Authorization: Bearer YOUR_API_KEY
6. Click Execute Workflow to test with a single file before enabling triggers

The gotcha most people hit first: n8n handles binary data differently in cloud vs. self-hosted. Cloud instances use an in-memory binary store; self-hosted defaults to filesystem storage with no size ceiling. If a workflow fails with "binary data not found" or silently drops file content, check Settings → Binary Data Mode before debugging anything else. Every template in this bundle explicitly sets the binary property name — data — to prevent the most common version of this failure.

---

Workflow 1: PDF → Text Extraction

This workflow receives a PDF via webhook, sends it to a conversion API, polls for completion, and returns clean extracted text with page count and confidence score — all within a single n8n execution, no intermediate file writes.

PDF extraction is harder than it looks. Scanned PDFs require OCR; generated PDFs need structure parsing; encrypted or rotated files break naive extraction entirely. Handling those cases in n8n with raw nodes requires multiple Code nodes and careful edge-case management. An API absorbs all of that.

Node chain:

Webhook (PDF upload) → HTTP Request (POST /v1/pdf/extract-text)
→ Wait (poll every 2 s, max 30 s) → HTTP Request (GET /v1/jobs/{id})
→ Code (shape output) → Respond to Webhook

HTTP Request POST configuration: - Method: POST - URL: https://api.convertfleet.com/v1/pdf/extract-text - Headers: Authorization: Bearer {{$credentials.apiKey}}, Content-Type: multipart/form-data - Body: file → binary from webhook input

Response shape:

{
  "text": "Full extracted content...",
  "pages": 12,
  "confidence": 0.97,
  "encoding": "UTF-8",
  "processing_ms": 840
}

Feed text directly into an OpenAI, Anthropic, or Pinecone node. No intermediate storage. In our experience building document pipelines, this single connection — PDF in, structured text out — replaces what used to require a dedicated Python microservice with its own deployment and monitoring overhead.

---

Workflow 2: DOCX → PDF Conversion

This workflow converts Word documents to print-ready PDFs with preserved layout — fonts, embedded images, margins — using a single API call, no LibreOffice container required.

LibreOffice headless in Docker works. It also needs a 500MB+ image, breaks on documents with custom fonts, and requires manual restarts when it hangs on complex files. For pipelines processing dozens of documents on a schedule, the maintenance cost compounds fast.

Webhook trigger config: - Path: docx-to-pdf - Binary property: data - Response mode: Respond to Webhook node (not immediate — we wait for conversion to complete)

The conversion request sends the DOCX as binary with options: {"compress_images": true}. An If node checks response.status === "success" before returning — corrupted files, password-protected documents, and malformed uploads all get a structured error response, not a silent failure that confuses the caller.

This is a good fit for: legal teams generating contracts from Word templates, finance teams exporting reports, any pipeline where the source is an editable document and the artifact is a shareable PDF.

---

Workflow 3: Image Batch Resize

This workflow processes multiple images in parallel, resizing each to target dimensions and optionally converting format, completing a batch of 50 images in roughly 20–25 seconds with five concurrent requests.

Sequential image processing in n8n is slow and memory-heavy. The template uses Split In Batches with concurrency set to five, which balances throughput against API rate limits. Adjust that number down to two or three if your API tier has tighter limits.

Two input modes: a webhook accepting a files[] array, and a scheduled trigger reading from a directory path (self-hosted only). The schedule trigger processes an upload folder overnight — nothing queues, nothing waits on a human.

---

Workflow 4: Audio Transcode

This workflow converts audio between formats with configurable sample rate, bitrate, and channel layout — designed specifically for pipelines that feed speech-to-text APIs with strict input requirements.

Sample rate mismatch causes pitch shifts. Stereo-to-mono conversion halves file size for STT inputs. Bitrate changes affect file size linearly but audio quality is more complex. All of these parameters are exposed in the Set node so you can tune without touching the HTTP Request configuration.

Set node configuration for STT prep:

{
  "output_format": "wav",
  "sample_rate": 16000,
  "channels": 1,
  "bitrate": "256k"
}

Whisper, AssemblyAI, and Deepgram all prefer 16 kHz mono WAV. Set those values and route the output directly to your transcription node — no intermediate storage required. For a 30-minute podcast episode, typical conversion time is 12–18 seconds at these settings.

---

Workflow 5: Bulk Multi-Format Conversion

This workflow reads a list of source files and target formats, routes each job to format-specific conversion logic via a Switch node, and collects all results into one structured output — handling a mixed-format list in a single execution.

This is the orchestrator. It doesn't convert files itself — it routes and collects. The Merge node runs in "Wait for all inputs" mode, so results arrive in a consistent structure even when individual jobs complete out of order.

Input JSON:

{
  "jobs": [
    {"file_url": "https://...", "target": "pdf"},
    {"file_url": "https://...", "target": "txt"},
    {"file_url": "https://...", "target": "jpg", "resize": {"width": 800}}
  ]
}

Real-world use case: content migration projects often need one source in three or four formats — a Word document becomes a PDF for distribution, TXT for indexing, and JPEG cover for a listing page. Manual tool-switching for 200 documents takes a full day. This workflow does it in an evening, unattended.

---

Connecting These Workflows to AI: n8n AI Automation Workflows

File conversion and AI processing are a natural pair — LLM summarization, embedding generation, and vision classification all require clean, correctly formatted input that raw uploads rarely provide. These five templates handle the conversion step so your AI nodes receive what they actually need.

This is the concrete shape of n8n AI automation workflows built on top of file conversion:

Document intelligence pipeline:

Webhook (raw upload) → PDF/DOCX → Text Extract
→ Anthropic / OpenAI node (summarize, classify, extract)
→ Pinecone / Supabase pgvector (store embedding)
→ Respond to Webhook

Image understanding pipeline:

Schedule trigger → Batch resize (standard dimensions)
→ OpenAI Vision / Gemini node (label, describe)
→ Airtable / Google Sheets (write results)

Audio transcription + summarization:

Webhook (audio upload) → Transcode (16 kHz mono WAV)
→ Whisper / AssemblyAI (transcribe)
→ Claude / GPT-4 (summarize, extract action items)
→ Slack / Email (deliver)

Each pipeline starts with one of these conversion templates. The conversion step is unglamorous infrastructure. But in our experience building document automation for teams, the single most common cause of AI pipeline failures is malformed or incompatible file input — incorrect encoding, wrong format, unexpected structure. Getting conversion right before the AI call eliminates the largest category of production errors.

n8n's native AI nodes (available from version 1.x onward) connect directly: the text output from Workflow 1 plugs into an Anthropic or OpenAI node's text input without a Code node intermediary. The resized binary from Workflow 3 feeds OpenAI Vision without additional transformation. Standard n8n data types all the way through.

---

Common Mistakes in n8n File Conversion Workflows

These are the failures that appear most often when teams build these pipelines without a reference template:

The self-hosted vs. cloud gap is real. Local n8n with 8 GB RAM handles 100 MB files fine. n8n Cloud has per-plan execution limits documented at n8n.io/pricing. Test with realistic file sizes before committing a production pipeline to the cloud tier — or self-host if your workloads are large.

---

Adapting These Templates for Your Stack

Swapping the conversion API means changing the URL in each HTTP Request node. Any service that accepts multipart file uploads and returns JSON works. Adjust the Wait node polling interval to match your API's typical processing time — some return in 500 ms, others take 30 seconds.

Common trigger substitutions: - Airtable / Notion record created → replace Webhook with the corresponding trigger node - S3 file upload → S3 Trigger → download → convert → upload to destination bucket - Email attachment → IMAP Trigger → extract binary → convert → reply or forward

For AI automation workflows, the output node varies: text goes to vector stores or LLM nodes; images go to vision APIs; audio goes to STT services. The conversion templates return standard n8n binary data and JSON, so the downstream AI node doesn't care which template produced the input.

---

Free Download

Grab the complete JSON bundle — no signup required:

• ⬇ n8n-workflow-templates-workflow-ece7c01c7eb077ab.json — All five workflows in one file. Import via Workflows → Import from File in n8n. Add your Convert Fleet API key in the Set or HTTP Request node, then execute. The bundle names each workflow so you can import selectively.

---

Frequently Asked Questions

What is an n8n workflow template and how do I import one? An n8n workflow template is a complete automation blueprint saved as a JSON file containing every node, connection, and configuration. Import it via the n8n interface (three-dot menu → Import → From JSON), paste the JSON, and click Import. The full workflow appears in your canvas ready for credentials and testing. No node-by-node setup required.

Can I use these n8n workflow examples without writing code? Yes. All five templates use standard n8n nodes — HTTP Request, Set, Switch, Merge, Split In Batches — and run without custom code. The Code nodes in PDF extraction and bulk conversion contain short JavaScript for reshaping JSON output. You can change inputs and outputs through the UI without touching those scripts.

Where can I find more n8n workflow templates and examples? The primary source is n8n.io/workflows, where the community shares automation templates across every category. GitHub repositories like zie619/n8n-workflows hold additional user-contributed examples. Search by node name or task ("PDF", "file conversion", "audio") to find starting points relevant to your use case.

How do I connect these templates to AI models for n8n AI automation workflows? The PDF→Text and DOCX→Text workflows return clean text in the text field of their JSON response. Connect that field directly to an OpenAI, Anthropic, or Ollama node's text input — no intermediary Code node needed. For image workflows, the resized binary output feeds vision API nodes directly. n8n's native AI nodes (1.x+) handle the data type automatically.

How do I handle large files — over 50 MB — in n8n workflows? Use an async callback pattern: trigger the conversion job, pass your webhook URL as the completion callback, and let n8n resume when the API calls back. This avoids synchronous execution timeouts. Self-hosted n8n has no enforced execution time limit; n8n Cloud limits vary by plan and are documented at n8n.io/pricing. Whatever tier you're on, test with your actual file sizes before going to production.

Are free file conversion APIs reliable enough for production workflows? Free tiers typically throttle at 50–200 requests per day and may watermark output files. Reliability varies — some services go down without notice, others maintain strong uptime even on free tiers. The Convert Fleet API offers consistent performance for evaluation without registration; paid tiers cover higher volume with uptime commitments. Test throughput and error behavior with realistic workloads before committing any pipeline to production.