Can Universal 3 Pro handle low-volume audio?

Yes. The speaker notes that even though the audio is low, the transcription still catches every word and is very accurate.

Does Universal 3 Pro properly identify and transcribe sound areas?

Yes, the video claims it identifies and transcribes the targeted sound areas really well.

Where can you try Universal 3 Pro?

The speaker invites viewers to head to their playground to try Universal 3 Pro.

What does Universal 3 Pro's prompting capability enable?

It expands use cases into fields that weren't previously possible, including non-standard audio files.

Universal-3 Pro transcribes ASMR

Q: What prompts can you use to capture specific sounds like whispering and tapping?

The video demonstrates adding a prompt to capture whispering, tapping, breathing, and mouth sounds, and have it transcribe those sounds as well.

AssemblyAI

Science & Technology3 min read2 min video

Feb 17, 2026|760 views|13

Save to Pod

Key Moments

TL;DR

Universal-3 Pro transcribes whispering ASMR accurately.

Key Insights

Whispering audio, even at low volume, can be transcribed accurately by Universal-3 Pro.

Prompts that specify whispering, tapping, breathing, and mouth sounds expand transcription utility.

The system identifies and transcribes audio cues (like tapping and breathing) in addition to words.

Transcription quality remains strong even with non-standard or low-volume audio content.

Demo encourages trying the tool in the playground to explore capabilities.

INTRODUCTION: ASMR AND TRANSLATION CHALLENGES

ASMR content is typically whispered or softly spoken, which can create transcription challenges for many systems. In this video, the presenter openly shares discomfort with ASMR yet uses it to illustrate Universal-3 Pro's transcription capabilities. The core claim is that the model can convert whispered words into readable text with high fidelity, even when the audio is quiet or unconventional. This opening frames the demo as both a personal observation and a demonstration of robustness under unusual acoustic conditions.

DEMONSTRATING ACCURATE TRANSCRIPTION AT LOW VOLUME

The demonstration centers on lowering the volume of an ASMR clip while maintaining accurate transcription. The presenter emphasizes that Universal-3 Pro transcribes every word despite the quiet input, proving that performance hinges on spoken content rather than loudness alone. This segment highlights practical reliability for low-volume or soft-spoken material, suggesting the model’s transcription is resilient to volume fluctuations and not dependent on loud cues to function effectively.

PROMPTING CAPABILITIES AND AUDIO CUES

A key takeaway is the use of targeted prompts to guide the transcription process. By requesting the model to capture whispering, tapping, breathing, and mouth sounds, the output can annotate and emphasize specific audio events. This demonstrates a flexible interface that tailors transcripts to specialized needs, enabling applications in accessibility, content analysis, and research where precise event-level details matter, not just textual words.

EXPANDING USE CASES BEYOND NORMAL AUDIO

The presenter claims that Universal-3 Pro expands use cases into domains previously unreachable with standard audio processing, especially for files that aren’t 'normal.' Non-standard contexts like ASMR or ambient recordings can now be analyzed with greater specificity. This broadens the potential across disciplines, allowing researchers and creators to study speech in soft or noisy environments and to extract precise cues such as taps or breaths alongside spoken content.

IMPLICATIONS FOR CONTENT CREATORS AND RESEARCH

In practical terms, these capabilities can streamline note-taking for podcasts or educational content, improve accessibility via richer transcripts, and enable granular indexing of audio events. The video suggests workflows where users upload a file, apply a targeted prompt, and receive a transcript capturing both words and defined sound events. These capabilities encourage iterative testing and refinement to achieve outputs that match specific project needs.

CALL TO ACTION: TRY IT IN PLAYGROUND

The video concludes with an invitation to experiment in Universal-3 Pro’s playground, providing a concrete path for users to reproduce the demo and explore new possibilities. By grounding the example in a real-world scenario with whispering and defined audio cues, the presenter offers a clear blueprint for users to experiment and adapt. The overarching message emphasizes accessibility and potential, inviting creators and researchers to discover how prompting can unlock advanced transcription for unusual audio content.

Mentioned in This Episode

●Tools & Products

Prompts cheat sheet for Universal 3 Pro

Practical takeaways from this episode

Do This

Upload whispering audio and use prompts to capture whispering, tapping, breathing, mouth sounds.

Lower the volume on quiet recordings and still rely on transcription.

Test prompting in the playground to verify results.

Avoid This

Don't ignore the potential of targeted prompts for unusual audio.

Don't assume low-volume audio cannot be transcribed without testing.

Common Questions

It’s described as able to transcribe whispering audio accurately, even when the volume is low; the video claims it transcribes that audio perfectly and catches every word.