Step 2: Configure Workflow Constants and File Chunking

Define the chunk size and FileFlows API settings, then split the incoming audio into 4MiB chunks for upload.

1. Open Set Workflow Config and set chunk_size to {{ 4 * 1024 * 1024 }}.
2. Set fileflows_url to your FileFlows base URL, e.g., http://0.0.0.0:5000.
3. Set flowUid to your FileFlows flow ID, replacing [YOUR_ID].
4. In Create 4MiB Segments, keep the provided JavaScript to read the binary upload and generate chunk metadata and binary parts.
5. Ensure Iterate Chunk Batches is connected after Create 4MiB Segments to control chunk uploads.

Step 3: Connect FileFlows Upload and Split Requests

Upload each chunk to FileFlows, validate the response, and trigger the server-side split operation.

1. In Upload Segment Part, set URL to ={{ $('Set Workflow Config').item.json.fileflows_url }}/api/library-file/upload and Method to POST.
2. Set Content Type to multipart-form-data and map body parameters: fileName to {{$json["fileName"]}}, chunkNumber to {{$json["chunkNumber"]}}, totalChunks to {{$json["totalChunks"]}}, and file to binary field chunk.
3. Use Exclude Temp Records to filter out temporary items with condition {{ $json.data }} notEndsWith .temp.
4. In Success Check, confirm the condition {{ $json.data }} exists to route valid uploads to Initiate Audio Split.
5. Configure Initiate Audio Split with URL ={{ $('Set Workflow Config').item.json.fileflows_url }}/api/library-file/manually-add and JSON body { "FlowUid": "{{ $('Set Workflow Config').first().json.flowUid }}", "Files": [ "{{ $json.data }}" ], "CustomVariables": { "callbackUrl": "{{$execution.resumeUrl}}" } }.
6. Keep Pause for Callback set to Resume webhook, HTTP Method POST, and Resume Amount 30 minutes to wait for FileFlows to send back split audio.

Step 4: Set Up Audio Extraction and Transcription Loop

Transform the callback binaries into items, loop through each audio part, and send them to OpenAI for transcription.

1. In Extract Audio Parts, keep the provided JavaScript to convert all returned binaries into individual items with a binary field named Audio.
2. Ensure Process Segment Loop follows Extract Audio Parts to iterate each segment.
3. Set up OpenAI Transcribe with Resource audio, Operation transcribe, and Binary Property Name Audio. Keep Language set to fr under options if you want French transcription.
4. Credential Required: Connect your openAiApi credentials in OpenAI Transcribe.
5. Use Throttle Pause after OpenAI Transcribe to control request pacing before looping back to Process Segment Loop.

Step 5: Combine Transcripts and Deliver Results

Aggregate all transcription pieces into a single text file and email it to the requester.

1. In Combine Transcripts, keep the JavaScript that concatenates item.json.text into a single transcription field.
2. Configure Build Text File with Operation toText and Source Property transcription.
3. Set Build Text File options to File Name transcription.txt and Encoding utf8.
4. In Email Transcript Delivery, set Send To to ={{ $('Incoming Form Capture').first().json.email }}, Subject to Your transcription is ready, and include the provided message.
5. Credential Required: Connect your gmailOAuth2 credentials in Email Transcript Delivery.

Step 6: Add Error Handling for Split and Transcription Failures

Ensure users receive feedback when FileFlows or transcription fails.

1. From Success Check, verify the false branch routes to Split Error Email.
2. In Split Error Email, keep Send To set to ={{ $('Incoming Form Capture').first().json.email }} and confirm the error message content.
3. From OpenAI Transcribe, ensure the error output connects to Email Error Notice to notify transcription issues.
4. Credential Required: Connect your gmailOAuth2 credentials in both Email Error Notice and Split Error Email.

Step 7: Test and Activate Your Workflow

Validate the full pipeline with a test submission before enabling it in production.

1. Click Execute Workflow and submit the Incoming Form Capture form with a small .mp3 file and a valid email.
2. Verify that Upload Segment Part, Initiate Audio Split, and Pause for Callback execute without errors.
3. Confirm OpenAI Transcribe produces text outputs and Combine Transcripts creates the transcription field.
4. Check that Email Transcript Delivery sends an email with transcription.txt attached to the provided address.
5. When testing is successful, toggle the workflow to Active to accept live form submissions.