Generate High-Quality Audio with Voxtral Small 24B 2507

Description

Generate High-Quality Audio with Voxtral Small 24B 2507

This workflow integrates the notdaniel/voxtral-small-24b-2507 model from Replicate to generate audio content from provided inputs. It handles API authentication, creates predictions, polls until completion, and outputs the final generated audio file.

⚡ Section 1: Trigger & Authentication

🔘 On clicking 'execute'** → Manually starts the workflow.
🔑 Set API Key** → Stores your Replicate API key to authenticate requests.

Benefit: Securely connects your workflow to Replicate’s API and ensures only authorized requests are made.

🎛️ Section 2: Create Prediction

🌐 Create Prediction** → Sends a request to Replicate’s API with parameters like:

audio: Input audio file (e.g., a reference sample).
max\_new\_tokens: Maximum number of tokens to generate (controls audio length/complexity).

Benefit: Starts the audio generation process with configurable input and settings.

⏳ Section 3: Polling & Status Tracking

🆔 Extract Prediction ID** → Captures the unique prediction ID and endpoint for polling.
⏱️ Wait** → Pauses for 2 seconds before re-checking.
📡 Check Prediction Status** → Polls Replicate’s API to see if the audio generation is done.
✅ Check If Complete** →

If finished: moves forward to process results.
If not: loops back to wait and check again.

Benefit: Efficiently manages asynchronous audio generation, ensuring the workflow only proceeds when results are ready.

🎧 Section 4: Process Result

📝 Process Result** → Extracts and structures final output data:

status (success or failure)
output (raw response)
metrics (generation statistics)
timestamps (created and completed times)
audio\_url (final generated audio link)

Benefit: Provides a clean, structured output that can be used in follow-up automations (e.g., sending audio to users, storing in a database, or sharing via email).

📊 Workflow Overview

| Section | Purpose | Key Nodes | Benefit |
| --------------------------- | ------------------------------- | ----------------------------------------------------------------------- | -------------------------------- |
| ⚡ Trigger & Authentication | Start workflow & authenticate | Manual Trigger, Set API Key | Secure execution |
| 🎛️ Create Prediction | Submit audio generation request | Create Prediction | Start model processing |
| ⏳ Polling & Status Tracking | Monitor prediction progress | Extract Prediction ID, Wait, Check Prediction Status, Check If Complete | Ensures reliable completion |
| 🎧 Process Result | Format and deliver output | Process Result | Clean audio result ready for use |

✅ Final Benefits

🔒 Secure authentication with Replicate
🎛️ Flexible audio generation using voxtral-small-24b-2507
⏳ Reliable polling until results are ready
🎧 Clean and structured audio output