<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Aayush on Medium]]></title>
        <description><![CDATA[Stories by Aayush on Medium]]></description>
        <link>https://medium.com/@aayushkumarjvs?source=rss-b6561abd8e6f------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Stories by Aayush on Medium</title>
            <link>https://medium.com/@aayushkumarjvs?source=rss-b6561abd8e6f------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 20:23:41 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@aayushkumarjvs/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Spending the summer in GSoC with Red Hen Lab]]></title>
            <link>https://medium.com/@aayushkumarjvs/spending-the-summer-in-gsoc-with-red-hen-lab-ad49e2c5b9d0?source=rss-b6561abd8e6f------2</link>
            <guid isPermaLink="false">https://medium.com/p/ad49e2c5b9d0</guid>
            <dc:creator><![CDATA[Aayush]]></dc:creator>
            <pubDate>Sun, 18 Oct 2026 04:28:52 GMT</pubDate>
            <atom:updated>2026-10-21T21:40:19.108Z</atom:updated>
            <content:encoded><![CDATA[<p><em>The progress of this Blog will be updated accordingly with the completion of the project</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ksUxMNu9VP36rXShjlBejA.png" /></figure><p><strong>Project Details: </strong><a href="https://summerofcode.withgoogle.com/proposals/details/wiZr5aCu"><strong>Red Hen Audio Tagger</strong></a></p><p>I am thrilled to announce that my proposal for the Google Summer of Code program has been accepted! It’s an incredible feeling to have the opportunity to work on real-world projects and gain valuable experience in the field. After submitting my proposal, I waited anxiously for the result. When I received the acceptance email, I felt a wave of excitement and a sense of responsibility. I was given the opportunity to work on a project that could make an impact in the community, and I was determined to make the most of it.</p><p>I am incredibly thrilled to be a part of Google Summer of Code 2023 because I have been participating in open-source projects since I began university. More information about me is available in <a href="https://aayushkumarjvs.github.io/">my portfolio</a>. I will be working for Red Hen Labs as an Open-Source Developer, and this blog will serve as a record of my weekly progress on the project I have proposed working on. Please click <a href="https://summerofcode.withgoogle.com/proposals/details/wiZr5aCu">this link</a> to access the project summary found on the Google Summer of Code website.</p><p>After completing my action item as a contributor in the Red Hen Audio Tagger project, I gained significant practical experience and insights into audio analysis, machine learning, and big data processing. I particularly appreciated the opportunity to work with advanced technologies such as TensorFlow, PyTorch, Flask, Docker, and more in a real-world context.</p><h3>Building an Audio Tagging Pipeline with Red Hen Lab</h3><p>During the summer of 2023, I participated in <strong>Google Summer of Code (GSoC)</strong> with <strong>Red Hen Lab</strong>, where I worked on developing an audio segmentation and tagging pipeline. The goal was to prepare audio and text data from video recordings so they could be used with End-to-End Automatic Speech Recognition (E2E ASR) systems.</p><p>This project gave me hands-on experience working with multimedia data, audio processing, Python scripting, and speech recognition workflows.</p><h3>GSoC 2023 Project Timeline</h3><h3>Community Bonding Period</h3><p><strong>Week 1 (May 23 — May 28):</strong> Mentors familiarized me with Red Hen Lab’s resources like Gallina, project requirements, and existing VRT data-processing workflows.</p><p><strong>Week 2 (May 29 — June 4):</strong> Explored the VRT file format and planned how to extract sentence text with corresponding start and end timestamps.</p><p><strong>Week 3 (June 5 — June 12):</strong> Designed the parsing workflow to generate segments.txt containing sentence text and timestamps.</p><h3>Phase 1 Evaluation</h3><p><strong>Week 4 (June 13 — June 18):</strong> Implemented the VRT parsing logic to extract sentence-level text and save it in segments.txt.</p><p><strong>Week 5 (June 19 — June 25):</strong> Developed the audio segmentation script to crop audio clips using sentence start and end timestamps.</p><p><strong>Week 6 (June 26 — July 2):</strong> Organized extracted clips in the audiofiles/ directory and generated samples.json containing audio paths, text, and durations.</p><h3>Final Evaluation</h3><p><strong>Week 7 (July 3 — July 9):</strong> Integrated VRT parsing, audio extraction, and manifest generation into a unified data-preparation pipeline.</p><p><strong>Week 8 (July 10 — July 16):</strong> Tested segments.txt, the extracted audio clips, and samples.json to check the consistency of the prepared data.</p><p><strong>Week 9 (July 17 — July 23):</strong> Refined the workflow for preparing sentence-level audio samples for downstream speech recognition and transcription.</p><p><strong>Week 10 (July 24 — July 31):</strong> Explored how the generated manifests could support an End-to-End Automatic Speech Recognition (E2E ASR) workflow.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/661/0*4zsNV-jypg164FgA" /></figure><p><strong>Week 11 (August 1 — August 6):</strong> Reviewed the data parsing and splitting workflow using the <a href="https://gallo.case.edu/snippet_cache/">Red Hen Lab snippet resource</a>.</p><p><strong>Week 12 (August 7 — August 13):</strong> Investigated pretrained ASR models and the manifest requirements for model training and evaluation.</p><p><strong>Week 13 (August 14 — August 20):</strong> Explored NVIDIA NeMo’s <a href="https://github.com/NVIDIA/NeMo/blob/main/tutorials/asr/ASR_CTC_Language_Finetuning.ipynb">CTC-based ASR fine-tuning tutorial</a> as a potential extension of the pipeline.</p><p><strong>Week 14 (August 21 — August 27):</strong> Consolidated the implementation details and documented the pipeline outputs, data format, and potential ASR integration.</p><p><strong>Week 15 (August 28 — September 5):</strong> Completed the project documentation and summarized the workflow for preparing sentence-level audio data for speech recognition and audio/video tagging.</p><h3><strong>What I learned</strong></h3><p>One key takeaway was understanding the practical challenges in developing and fine-tuning machine learning models for audio tagging, and the importance of accuracy and relevance in the tagging process. I’ve learned the art of managing complexity in a vast project that involves numerous technologies and areas like machine learning, deep learning, signal processing, and audio analysis.</p><p>Furthermore, participating in GSOC under Red Hen Labs has broadened my perspective on the impact of our work. Seeing our project applied in the field of humanities underscored the fact that the technologies we develop can have wide-ranging implications beyond the tech world, underscoring the societal impact of our work.</p><p><strong>CONSOLIDATED REPORT: Red Hen Audio Tagger V1.0 </strong><a href="https://docs.google.com/document/d/1nM8u4aTmk_2o3C3bAXuIUPANOhB7KQbr66c7Xe76OLs/edit?usp=sharing">https://docs.google.com/document/d/1nM8u4aTmk_2o3C3bAXuIUPANOhB7KQbr66c7Xe76OLs/edit?usp=sharing</a></p><p><strong>Mentors: </strong><a href="https://www.linkedin.com/in/sabyasachighosal/"><strong>Sabyasachi Ghosal</strong></a><strong>, </strong><a href="https://www.linkedin.com/in/karan-singla-a41155109"><strong>Karan Singla</strong></a><strong>, </strong><a href="https://en.wikipedia.org/wiki/Mark_Turner_(cognitive_scientist)"><strong>Mark Turner</strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ad49e2c5b9d0" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>