In this guide
Inspect the original file’s header to learn its stored format. Read encoding, container bits, and valid bits together. Those values help with compatibility and size planning, but they cannot certify sound quality or tell you how the source was originally recorded.
To check a WAV file’s sample rate, bit depth, and channels, inspect the original file rather than relying on its name or an editor’s export preset. Open the free WAV file inspector, choose a supported recording, and read the header report. The file stays in your browser and is not decoded or converted.
This check answers a practical question: what format is stored in this file right now? That is useful when preparing an upload, delivering audio to a collaborator, or checking whether a recent export matches your intended settings. It does not answer whether the recording sounds good.
Read the report in this order
- Encoding. Establish whether the samples use integer PCM or IEEE floating-point storage.
- Sample rate. Read the stored setting in Hz, such as 44,100 or 48,000.
- Channels. Check whether the file stores one channel, two channels, or more.
- Container and valid bits. Read these alongside the encoding, especially for an extensible WAV.
- Duration and size. Compare the declared data duration and actual selected file size with your project or upload requirements.
Save the original and record its provenance if that matters to the project. A file name such as “interview-48k-24bit.wav” is a convenient label, but the label can survive an export made with entirely different settings. The report reads format fields rather than trusting the label.
What does the sample rate tell you?
The sample rate describes the number of sample frames per second. A frame contains a stored sample for each channel, so a stereo file has two channel samples at each frame. Microsoft’s WAVEFORMATEX documentation defines the sample-rate and channel fields, along with the byte rate and frame alignment used for uncompressed audio.
Check the selected file after export. An editor’s project rate or playback setting is not necessarily the rate written in a particular source file. This inspector reads the original WAV header directly, so choosing a 44,100 Hz file produces a 44,100 Hz report without a playback context resampling it first.
Raising an export rate does not establish that the microphone originally captured additional detail. Keep the source history separate from the current format. If a collaborator asks for a delivery setting, make that export deliberately and check the resulting file rather than changing settings to chase a larger number.
For a short supported mono or stereo recording, the free 44.1/48 kHz sample rate converter can make a 16-bit PCM WAV at either rate. It resamples the samples rather than relabeling the header. Inspect its result as a new file; its rate, bit depth, metadata, and exact decoded frame count need not match the source.
Container bits, valid bits, and float are different ideas
For integer PCM, the container width tells you how many bits are used to store each channel sample. An extensible header can separately declare how many of those bits are valid. A file with 24 valid bits in a 32-bit container uses four bytes per channel sample, even though the declared signal precision is 24 bits.
Microsoft’s WAVEFORMATEXTENSIBLE reference explains that the container width is a multiple of eight and that valid bits can be fewer. The inspector shows both fields instead of calling either one a quality score. If an extensible valid-bit field is zero, the report says it is not specified.
Read the encoding too. A 32-bit float WAV and a 32-bit integer PCM WAV use different sample representations. Basic float headers do not carry a separate valid-bit declaration, so that row is shown as not specified. Do not interpret a float container size as an equivalent number of integer precision bits.
A large container can hold audio derived from a smaller or lossy source. The header cannot establish the effective detail of that source, certify an untouched recording, or detect an MP3 history. Keep the best original and avoid treating a format conversion as evidence of restored information.
Mono and stereo describe storage, not stems
A mono WAV stores one channel; a stereo WAV stores two. Neither number tells you how many people or instruments were recorded. One mono channel can contain a full band, and two stereo channels can both contain the same mixture of voices and music.
Use the channels and stems guide to decide whether extracting an existing side solves your problem. If a recorder already placed separate microphones on left and right, the free channel splitter may provide the audio you need. If the source you want overlaps other sounds on both sides, channel count alone does not separate it.
For multichannel files, speaker positions are another piece of information. The inspector counts stored channels and notes an unspecified or inconsistent extensible speaker mask. It does not turn a channel count into a surround layout, label speakers automatically, or preview a downmix.
A worked WAV size and duration example
Consider a basic one-second stereo integer PCM file at 44,100 Hz with a 24-bit container. Each frame stores two three-byte samples, or six bytes. Its 44,100 frames occupy 264,600 audio data bytes. With a basic 44-byte header and no additional chunks, the complete file is 264,644 bytes.
| Stored format | Bytes per frame | Audio data | Complete file |
|---|---|---|---|
| 44,100 Hz · stereo · 24-bit integer PCM | 6 | 264,600 bytes | 264,644 bytes with a basic header |
| 48,000 Hz · mono · 16-bit integer PCM | 2 | 96,000 bytes | 96,044 bytes with a basic header |
| 48,000 Hz · six channels · 24 valid bits in a 32-bit container | 24 | 1,152,000 bytes | 1,152,068 bytes with a basic extensible header |
These examples show how the stored format affects file size. The last example uses its 32-bit container width for storage; substituting 24 valid bits would undercount its bytes.
For supported uncompressed WAV, the inspector calculates duration as audio data bytes divided by bytes per frame, then divided by sample rate. Metadata and padding do not count as sample frames. A file can therefore be larger than a simple audio payload estimate while keeping the same audio duration.
Microsoft’s RIFF overview describes the format and data chunks, their sizes, and padding. WAV headers are not always 44 bytes long. The inspector skips supported extra chunks, checks their bounds, and includes every selected-file byte in the actual-size report.
Use actual size when preparing an upload
Saturalabs’ AI upload limit is 52,428,800 bytes, or 50 MiB. A displayed size in decimal MB is a different unit. The inspector shows both and compares the actual selected file size with the byte limit, including headers, metadata, and any bytes after the declared RIFF container.
That verdict checks size only. It is not a promise that the uploader, a browser player, or another editor supports this particular codec and layout. If the recording is too large, prepare a representative excerpt in your editor. The size calculator helps plan a common mono or stereo PCM export; the free trimmer handles supported short mono/stereo MP3 and WAV inputs.
The trimmer and channel splitter make their own 48 kHz, 16-bit PCM exports. Keep that delivery format separate from the original values shown by the inspector. Follow the preparation guide to retain complete phrases and any original timeline offset before processing a clip.
What an unsupported or inconsistent result means
This inspector supports little-endian RIFF/WAVE with one direct data chunk: 8-, 16-, 24-, or 32-bit integer PCM and 32- or 64-bit float, including supported extensible headers. It does not support compressed WAV, RF64, BW64, big-endian RIFX, segmented audio, or more than 1,024 chunks.
An unsupported layout is a reason to inspect the file in your audio editor. It is not a finding that the recording sounds bad. For an incomplete or contradictory header, recover the intact source or make a new standard WAV export while keeping the original.
A successful header report is also limited: it does not inspect sample values, verify every part of the audio payload, measure loudness or clipping, or certify lip sync. Listen to the recording before an edit. For video, the audio replacement guide explains alignment and final-export checks.
Common questions
Can I check a WAV without uploading it?
Yes. The free inspector reads small sections of the selected file in your browser. It does not send audio to Saturalabs, start an isolation job, or require an account.
Is a 96 kHz or 32-bit file always the better source?
The header numbers alone cannot decide that. Source history, the actual recording, and the edit you need matter. Prefer an intact original and compare the sound rather than choosing by the largest settings.
Why is the file larger than the size calculator predicted?
The calculator assumes a basic header for its PCM estimate. Metadata, an extensible header, and other chunks add bytes. Compare actual size and audio data size in the inspector to see that difference.
Can the inspector tell whether audio is suitable for voice isolation?
No. It reports storage fields, not a separation score. After preparing a representative clip, use the voice isolation workflow to compare the requested voice with the original and inspect any lost speech before choosing a result.