High-Quality Audio Generation using Replicate Voxtral 24B - n8n Workflow

• Podcasters or content creators needing automated audio post-production.


• Developers looking for advanced n8n templates integrating external AI APIs.


• Users who need to implement robust asynchronous polling logic in an n8n workflow.


• Automation specialists deploying AI-powered n8n node integrations.

This powerful n8n workflow solves the complex challenge of managing asynchronous API calls, specifically for long-running AI generation tasks like high-fidelity audio creation. The workflow leverages the Voxtral Small 24B 2507 model hosted on Replicate. Since audio generation can take time, this n8n automation utilizes a sophisticated polling loop rather than timing out. When triggered, the n8n node first initiates the generation job, then continuously checks the status until the audio file is complete, ensuring reliable output retrieval. This specific n8n workflow provides a ready-to-use blueprint for handling advanced, complex AI integrations and demonstrates best practices for building an efficient n8n node polling system.

The entire process is designed for robust asynchronous handling within an n8n automation environment:

1. Trigger and Setup: The process begins with the Manual n8n trigger ('On clicking 'execute''). Following the activation of the n8n workflow, a Set n8n node configures the required Replicate API key.


2. Initiate Generation: The first HTTP Request n8n node calls the Replicate API to create a new prediction job using the notdaniel/voxtral-small-24b-2507 model.


3. Extraction: A Code n8n node extracts the unique Prediction ID and the initial status required for subsequent polling, essential data for the loop that follows in the n8n workflow.


4. Polling Loop (Wait): The workflow pauses using the Wait n8n node for a predefined period (2 seconds) before attempting to check the job status.


5. Status Check: Another HTTP Request n8n node uses the extracted Prediction URL to query Replicate for the job status.


6. Conditional Check: An If n8n node evaluates the status. If the status is "succeeded," the n8n workflow proceeds to process the result. If the status is anything else (e.g., "processing"), the If node routes the execution back to the Wait n8n node, initiating the next poll cycle.


7. Final Output: Once successful, the final Code n8n node processes the output, extracting the direct audio URL and key generation metrics, concluding the execution of this robust n8n workflow template.

1. Import the n8n workflow: Copy the provided JSON and import it directly into your n8n instance.


2. API Key Configuration: Locate the 'Set API Key' n8n node. Replace the placeholder YOURREPLICATEAPIKEY with your actual Replicate API key.


3. Credential Setup: Ensure the HTTP Request n8n node ('Create Prediction') is properly configured to use Generic HTTP Header Authentication, referencing the replicateapikey variable set in the previous n8n node.


4. Input Configuration: Modify the input parameters within the 'Create Prediction' n8n node's JSON body to specify the input audio URL and desired maxnew_tokens.


5. Execution: Run the n8n workflow using the Manual Trigger for testing, or set up a real n8n trigger (like a Webhook or Schedule) for production use.

On clicking 'execute' (Manual Trigger n8n trigger): The starting point for this specific n8n workflow, primarily used for testing and manual execution.
Set API Key (Set n8n node): Defines the replicateapikey used in the HTTP requests, crucial for authenticating the n8n workflow with Replicate.
Create Prediction (HTTP Request n8n node): Initiates the audio generation task. Key Configuration: Uses POST method to https://api.replicate.com/v1/predictions, specifying the notdaniel/voxtral-small-24b-2507 model version and Bearer authorization via the stored API key.
Extract Prediction ID (Code n8n node): Parses the initial response to capture the predictionId and constructs the polling URL for the subsequent n8n nodes.
Wait (Wait n8n node): Implements a two-second delay, managing the polling frequency to prevent rate limiting while waiting for the AI job to finish.
Check Prediction Status (HTTP Request n8n node): Polls the specific prediction URL to track job completion using the same Bearer authentication setup.
Check If Complete (If n8n node): The core logic that determines the workflow flow. It checks if the API response status is exactly "succeeded." If false, it loops back to the Wait n8n node; if true, it proceeds.
Process Result (Code n8n node): The final processing n8n node. Extracts the completed audio URL (result.output) and relevant metadata for easy consumption by downstream systems in the n8n workflow.