AWS Transcribe n8n integrationAutomate AWS Transcribe with n8n.
Where does the audio go once AWS Transcribe has run? The AWS Transcribe n8n integration exposes one resource, Transcription Job, with 4 operations behind it. No trigger node ships with this tool, so a Schedule or a Webhook step opens the workflow. Written for teams turning recordings into text they can search.
Verified Trustpilot reviews · AI, automation & growth agency
What does the AWS Transcribe n8n integration actually do?
It wires an n8n workflow into Amazon's speech to text service. The node is n8n-nodes-base.awsTranscribe, it sends data to AWS Transcribe, and it gives you a single resource, Transcription Job, with 4 operations: create a job, read one back, list them, delete one. Everything around it, the file, the credential, the destination, is ordinary n8n. The node sits in the Development category of the catalog and runs on schema version 2.
Scenario one, a searchable archive of sales calls. A Schedule Trigger wakes the workflow, the recordings already sit in AWS S3, and transcriptionJob.create opens one job per file with Media File URI pointing at the S3 object. A later pass calls transcriptionJob.get with Return Transcript on, and the text lands in Google Sheets next to the call date.
Scenario two, a moderation pass on audio your users upload. Vocabulary Filter Name and Vocabulary Filter Method decide what happens to the words you flagged: mask swaps them for three asterisks, remove drops them without placeholder text, tag marks them in the output and leaves the wording alone. The tagged transcript goes to Slack so a human reads it before anything is published.
Scenario three, a nightly sweep for jobs that never finished. transcriptionJob.getAll filtered on Status FAILED returns them as one item each, and the list is written to Postgres so the same file is not submitted twice tomorrow morning.
The node covers the Transcription Job resource. Whatever falls outside those 4 operations goes through the HTTP Request node, which calls any AWS Transcribe endpoint while reusing the same AWS credential through predefined authentication. That is the normal route in n8n, not a hack around it.
Two published templates already pair AWS S3 with AWS Transcribe, one of them adding Google Sheets on top, so the storage plus transcription shape is well travelled. If n8n itself is new to you, the n8n review covers what the platform is good at, and the n8n training walks through building a workflow end to end.
How do you connect AWS Transcribe to n8n?
- 01
Pick the credential type
n8n offers two credential types for AWS and both authenticate this node. AWS (IAM) signs requests with an access key. AWS (Assume Role) assumes an IAM role through AWS STS, and the credentials used for the
AssumeRolecall are either typed in by hand or read automatically from the environment n8n runs in. Self hosting n8n on EKS, ECS or EC2 makes the second one attractive, since no static key ends up stored in n8n. - 02
Generate the access key in IAM
Open the IAM console, select your user name in the upper right, then Security credentials. In the Access keys section choose Create access key, pick your use case or Other if none is offered, and set a description tag such as
n8n integration. Reveal the Access Key ID and the Secret Access Key, then paste both into the n8n credential. The AWS CLI and the AWS API generate keys too if you prefer that route. - 03
Save it once, reuse it everywhere
The credential also asks for your AWS Region, so select the one your transcription jobs belong to. Save it from the Credentials menu and every workflow can pick it from the node dropdown, no retyping. For a short lived key, turn on the temporary security credential option and add the Session token alongside the key pair. The same credential then serves the other AWS nodes in your instance.
What can the AWS Transcribe node do?
The AWS Transcribe node exposes 4 operations across 1 resources. For each one: the node as you configure it in n8n, the required fields, and our field notes.
| Resource | Create | Get | Get Many | Delete |
|---|---|---|---|---|
| Transcription Job | ✓ | ✓ | ✓ | ✓ |
Transcription Job
4 operationsCreate a transcription job
transcriptionJob.createWhat you see in n8n
Notes & use cases
Submits a media file to AWS Transcribe and returns the job it opened.
Key parameters
- Media File URI: the S3 object location of the input media file.
- Job Name: the name of the job, often
{{ $json.key }}, since you reuse it to fetch the transcript. - Language: the language used in the media, picked from
en-US,en-GB,de-DE,en-IN,en-IE,ru-RUores-ES, unless Detect Language identifies it for you. - Options: 6 more fields, among them Max Speaker Labels and Channel Identification, which cannot both be set in one request without a BadRequestException.
Delete a transcription job
transcriptionJob.deleteWhat you see in n8n
Notes & use cases
Removes a job from AWS Transcribe, which also takes it out of what Get many transcription jobs returns afterwards.
Key parameters
- Job Name: the name of the job, and the only thing this operation reads, so it has to match exactly the name Create a transcription job registered.
Get a transcription job
transcriptionJob.getWhat you see in n8n
Notes & use cases
Reads one job back. By default the response carries metadata about the transcript, not the transcript.
Key parameters
- Job Name: names the job to read.
- Return Transcript: enable it to retrieve the transcript itself rather than the metadata around it.
- Simplify: returns a simplified version of the response instead of the raw data, which keeps the expressions downstream short.
Get many transcription jobs
transcriptionJob.getAllWhat you see in n8n
Notes & use cases
Lists the transcription jobs the credential can see, one n8n item per job.
Key parameters
- Return All: on, n8n walks the API pages and brings back every job; off, Limit caps how many come back.
- Job Name Contains: keeps only the jobs whose name contains that string, useful when the name encodes a team or a date.
- Status: narrows the list to
COMPLETED,FAILED,IN_PROGRESSorQUEUED.
IN_PROGRESS and posts a single line to the ops channel instead of one alert per job.Need help automating AWS Transcribe with n8n?
A person reads every message.