awsPractitioner

Transforming Sound Design in Your IDE with Mirelo AI and Kiro

5 min read AWS DevOps BlogOct 1, 2026Reviewed for accuracy

Practitioner — Hands-on experience recommended

In the fast-paced world of software development, the ability to create sound effects on the fly can significantly enhance user experience. Mirelo AI addresses this need by allowing developers to generate high-quality audio directly from their Integrated Development Environment (IDE). By leveraging the Model Context Protocol (MCP), Mirelo AI transforms text prompts or video clips into production-ready sound effects, streamlining the sound design process.

Mirelo has built a hosted server on the MCP, which acts as a bridge between your IDE and their powerful sound generation models. When you mention sound, audio, or effects in your prompt, Kiro activates and connects to Mirelo's server. You simply describe the sound you want, and Kiro uses the MCP to call Mirelo's models, returning a finished audio file right into your project. You can specify parameters like the sound description in natural language, duration in milliseconds, and the desired output format, making it flexible for various use cases.

While this integration is powerful, remember that the provided code samples are for demonstration purposes only. Ensure you store API keys securely and implement error handling before deploying in a production environment. Additionally, always test thoroughly to avoid unexpected failures during sound generation.

Key takeaways

→Utilize the Model Context Protocol (MCP) to connect your IDE with Mirelo AI for sound generation.
→Specify sound parameters like prompt, duration_ms, and output_format for tailored audio results.
→Implement proper security measures for API keys and include error handling in your production code.

Why it matters

Integrating sound design directly into your IDE can dramatically reduce development time and improve the quality of user interactions, leading to a more engaging product.

Code examples

Bash

1# 1. Submit the job
2JOB=$(curl -s https://api.mirelo.ai/v2/text-to-sfx/v1.6/jobs \
3  --request POST \
4  --header 'Authorization: Bearer sk-<your-api-key>' \
5  --header 'Content-Type: application/json' \
6  --data '{
7    "prompt": "Heavy rain on a metal roof with distant thunder",
8    "duration_ms": 45000,
9    "output_format": "mp3"
10  }' | jq -r '.job_id')
11
12# 2. Poll until the job leaves the "processing" state
13while true; do
14  STATUS=$(curl -s "https://api.mirelo.ai/v2/text-to-sfx/v1.6/jobs/$JOB" \
15    --header 'Authorization: Bearer sk-<your-api-key>' | jq -r '.status')
16  [ "$STATUS" = "succeeded" ] && break
17  [ "$STATUS" = "failed" ] && echo "generation failed" && exit 1
18  sleep 3
19done
20
21# 3. Fetch the result URL and download the clip
22curl -s "https://api.mirelo.ai/v2/text-to-sfx/v1.6/jobs/$JOB" \
23  --header 'Authorization: Bearer sk-<your-api-key>' \
24  | jq -r '.result_urls[0]' \
25  | xargs curl -o rain.mp3