Audiogram Design and Waveforms

Use the Audiogram Maker to try these decisions with local audio. This guide explains the peak display and the design controls without treating them as transcription, frequency analysis, or loudness certification.

Give the design one listening task

Picture a small listening workshop preparing an exercise about everyday sounds. In this hypothetical example, the recording contains a paper sheet unfolding, a pause, and two light taps. A presenter says: The first tap is our marker. Listen to the paper unfold, then notice the pause before the second tap. The intended video is not a sound identification test scored by software. It is a simple invitation to pay attention. A useful design names that task, provides the presenter context, and lets the actual recording carry the detail. The moving bars can accompany the exercise without pretending to understand its events.

Choose the audience action before selecting a palette. Here the action is to notice the pause, not to admire the loudest transient or memorize the presenter name. That suggests a headline such as Listen for the space between sounds. A long technical title would work against the purpose of the exercise. For another recording, the action might be understanding a repair step or hearing a speaker qualification. Different actions deserve different wording even when the same waveform renderer is used. A reusable visual structure is useful; a repeated message that ignores the recording is not a finished design.

Start with an existing local audio only file and keep your original. The maker does not record a microphone, extract a soundtrack from video, or retrieve an audio URL. Supported preparation requires a source no larger than 32 MiB, no longer than 180 seconds, and mono or stereo, with actual codec decoding available in the browser. Select a continuous one to sixty second interval that contains the complete listening exercise. Load example uses original synthesized tones to demonstrate the controls. Those tones can help you inspect motion, but they cannot establish whether a spoken instruction or a subtle paper sound remains clear.

Understand the sample peaks behind the bars

After preparation, the working sound is PCM audio at 48 kHz. PCM represents a changing waveform as a sequence of numerical samples. A sample peak is the largest absolute sample value within a chosen slice of that sequence. Taking the absolute value treats a negative swing as activity just like a positive swing. The maker prepares peak information in one hundred time bins per second. At 48 kHz, a hundredth of a second contains 480 sample frames, so a short transient can affect a bin even though the visual display is much simpler than the original sample sequence.

With stereo sound, the peak view takes the greatest absolute sample value across the channels. It does not show two independent channel traces, and it does not add the left and right channels into a claim about spatial position. The output audio itself retains mono or stereo. A tap recorded strongly on one side can therefore produce a visible peak without telling the viewer where that tap occurred. For the paper exercise, use the bars as a compact activity cue. Do not label one side of the graphic as the presenter and the other as the paper unless your own separate explanation supplies that meaning.

The height scale is linear relative to full scale, rather than an automatic attempt to make every clip fill the available height. A quieter recording may show smaller movement. Increasing gain can enlarge peaks because it changes the selected processed audio, but a small graphic is not by itself a reason to make the sound louder. This is a time-based amplitude summary, not a frequency spectrum. It cannot identify a pitch, separate rustling from speech, recognize words, or measure emotional emphasis. Similar peak shapes can come from very different sources, so the authored headline still has to explain what listeners are hearing.

Read the centered window without mistaking it for captions

Forty eight displayed bars cover a two second window centered on the current video time. At a video time of 3.5 seconds, that window covers the neighborhood from 2.5 to 4.5 seconds of the selected audio. Part of the visible window therefore lies before the exact playback position and part lies after it. This is intentional context around a moment, not a claim that every visible bar is sounding at that instant. In the workshop exercise, the approach of the second tap may be represented in the nearby window before the tap is actually heard during playback.

The one hundred bins per second describe the prepared peak data, while forty eight describes the number of bars drawn at once. Those are different quantities; one visible bar should not be explained as one word, one syllable, or one original audio sample. The video targets twenty four frames per second, which is another separate rate. The sound remains a continuous selected audio interval rather than a string of twenty four sound events each second. Keeping these roles distinct helps you explain why a compact animation follows a rich recording without being a literal drawing of every sample or a timestamped language map.

At the selected beginning and end, the centered window extends beyond the available sound. Those outside positions use zero, which can make the edge of the graphic look quieter without indicating a damaged file. During a real pause, background sound may still produce small peaks. A microphone bump may produce a taller bar than the presenter voice. Use Preview time to inspect these states, then confirm them in the actual video. Do not judge caption timing by when a bar becomes tall. Timed words must be checked against audible speech, because this display contains no word recognition or automatic linguistic alignment.

Build a clear hierarchy with four text roles

For the workshop, put Listen for the space between sounds in Headline and Everyday listening exercise in Small label. A verified presenter name can go in Byline, while Footer line might say Paper, pause, two taps. These are static single line fields. Their limits are eighty, forty, sixty, and sixty characters respectively. A field can be shorter or optional when the design does not need it. Before creating a video, make sure the headline is visible and meaningful. A draft with an empty headline can preserve settings, but it is not ready to render the exercise.

Do not ask all four fields to explain the same idea. The headline gives the task, the label identifies the kind of material, the byline identifies the speaker, and the footer supplies compact context. If the footer only restates the headline, remove or rewrite it. Long text is fitted to the canvas rather than deliberately shortened into an ellipsis, but fitting is not a reason to use every available character. A line that technically fits can still demand too much attention. Prefer a complete plain phrase over a dense list of workshop topics that competes with the sound.

Read the artwork at a reduced viewing size and without playing the recording. Can you identify the exercise immediately, and can you distinguish the speaker from the instruction? Then play the audio and ask whether the text adds useful context rather than duplicating a full transcript. The Thumbnail and Title Preview can be a separate wording review when you need to compare a compact title presentation. It does not automatically receive this project or predict the native destination layout. Keep the final judgment with the actual canvas, the real words, and the place where the video will be watched.

Change gain and fades for sound, not decoration

Level (dB) multiplies the selected audio amplitude. A six decibel increase is approximately a doubling of amplitude, not a promise that listeners will experience exactly twice the loudness. For example, a sample peak of 0.5 becomes just under 1.0 after a six decibel increase, while a slightly higher starting peak could exceed full scale. The maker permits gain from minus twenty four to plus twelve decibels but refuses selected processed peaks above full scale before AAC encoding. It does not hide the problem behind a limiter. Lower the level or improve the source when that check fails.

The waveform follows the processed selection, so gain and fades affect its peak data as well as the sound you create. A fade reduces the edge amplitude over a chosen interval; it does not remove a room noise throughout the recording. Both fades are linear, each may last zero to ten seconds, and they cannot overlap. On an eight second excerpt, a five second fade in plus a four second fade out is invalid because the regions overlap. A much shorter edge fade might suit the paper exercise, but only if it leaves the marker taps and the spoken instruction easy to hear.

Do not optimize the picture by pushing every peak toward the top. The tool does not perform compression, automatic normalization, denoising, LUFS measurement, or true peak analysis. Short sample peaks and perceived loudness describe different aspects of a recording. A brief tap can peak strongly while a quieter-looking bed of continuous noise feels tiring. AAC encoding is lossy and may alter peaks again after the preparation check. Use a steady listening volume, compare the original with the encoded video, and choose the least intervention that makes the exercise comfortable without changing the relationship between the sound and the silence.

Compare canvases and palettes with the same message

The three available canvases are Portrait at 720 by 1280 pixels, Square at 720 by 720, and Landscape at 1280 by 720. These are preparation sizes, not native platform limits or upload guarantees. For a workshop screen, landscape may give the instruction a comfortable horizontal reading path. A square version may suit a compact post, while portrait changes the overall balance around the same words. Compare all three with the complete byline and footer present. Do not judge the design only from a blank sample, because the real wording determines how much visual space the hierarchy needs.

The fixed palettes are Violet, Forest, Rose, Ocean, and Ink. Try Rose for the listening exercise, then compare Ocean if the text feels clearer in that combination. Palette choice can support the tone, but it should not imply that the bars classify sounds by mood or category. Their movement comes from the selected audio, not from a color-based analysis. There is no custom background image or logo upload. Work within the actual visual vocabulary: authored text, bars, elapsed and total time, and progress. A clear teaching clip does not need an invented photographic scene that the maker cannot produce.

Inspect an active moment, a quiet pause, the opening edge, and the ending edge with Preview time. The chosen slider position is a UI viewing aid only; it is not saved in the settings or used as a special video start point. The start and end controls define the audio interval instead. Once a composition reads well in several states, check its overall density at the size your audience is likely to see. A busy waveform should not become the main subject when the exercise asks people to notice a pause. Shortening a label can be more effective than repeatedly switching between colors.

Distinguish a design preview from an encoded result

Before Create video, you are inspecting a design preview. After creation, review the actual MP4 through manual playback. The output uses H.264 video and AAC audio, targeting twenty four frames per second with 128 kbps audio at 48 kHz and retaining mono or stereo. Required codec availability depends on the browser. If the requested audio path cannot work, a clear failure is better than a video that quietly loses the exercise sound. The original file is not changed by this process. Confirm the presenter cue and both marker taps in the encoded result before approving the design as a finished teaching asset.

The actual encoded first frame PNG and the encoded midpoint frame in the PDF are evidence of particular output frames, not substitutes for watching motion and hearing sound. Look for readable words in those frames, then play across the full pause and final tap. Requested interval length and MP4 container duration can differ slightly because the final video frame or AAC padding may extend timing. That small variation does not establish an audible fault by itself. Check the content. If the first syllable is missing or a tap sounds cut, revise the interval or fade and create a new version rather than accepting a plausible still image.

A waveform and a summary headline do not make speech accessible without sound. When you have supplied timed captions, review them against the final cut with the Subtitle Editor. Use the separate Add Subtitles to Video workflow when a captioned video is needed. These are deliberate file handoffs, not automatic project imports or transcription features of the audiogram maker. Keep meaningful sound descriptions in your caption work where appropriate to the exercise. If you revise the excerpt afterward, review timing again so the instruction and the audible markers still agree.

Save a reproducible recipe without confusing it with media

Download the MP4 as the playable deliverable and settings JSON as a separate recipe. The settings store design and audio adjustment choices, not audio bytes, video, the original filename, or review metrics. Keep the source in your own folder if you want to reproduce the workshop exercise later. Opening settings keeps whatever audio is currently loaded, so verify the interval against that source; a mismatched range is not silently clamped. Loading a valid new source preserves the visual design but resets timing, gain, and fades. Those are different operations, and neither proves that the right recording has been selected.

Any project edit makes earlier generated outputs stale. A change from two taps to three taps therefore needs a fresh video and another listen, not merely a new note attached to an old MP4. CSV, TXT, copied review, and PDF help document the approved version. The PDF includes an actual encoded midpoint image and complete settings and review attachments, without embedding the source audio or video. Its visible text and waveform can still reveal the exercise content. Inspect the report itself before sharing it, and send the MP4 separately when the recipient needs a playable teaching clip rather than a production record.

Save useful files before leaving the page. Page hide cleans up decoded buffers, generated results, and media URLs; Restore audio is an explicit action only when a cached file remains available. Reloading requires selecting the source again, so the open tab is not a durable project archive. Finish with one final listening question: does the design direct attention toward the intended pause without pretending to explain the sound automatically? Later publishing and supported account routines can be organized separately, including with SMTasker where relevant. This maker does not automatically host, caption, schedule, or publish the workshop video you have just created.

Questions about Design and Waveforms

Continue with another audiogram guide

Apply the idea in the Audiogram Maker, or continue with another audio-to-video guide. Browse more utilities in all free tools.