Exported Video Audio Pitch Changed, Speech Faster or Slower? Sample Rate, Sequence Settings and Tag Diagnosis Guide

2026-11-18 | GPU Repair | 40 minutes | HK GPU Fix Editorial Team

Exported Video Audio Pitch Changed, Speech Faster or Slower? Sample Rate, Sequence Settings and Tag Diagnosis Guide

Key Summary

An export whose voices sound higher or lower than they should, and faster or slower by the same amount, points at a sample rate label, not at the recording itself. This guide uses three comparisons — start against end, MediaInfo against sequence, one player against another — to separate a label fault from a sequence, player, or clock fault.

The Shape of the Problem: Sound Present, Pitch and Speed Wrong Together

This guide covers one specific shape, and the shape itself is the whole diagnosis. An edited video is exported, and the finished file has sound — the sound is definitely there — but the voices sound wrong. They are higher than they should be, like a cartoon voice, and the speech runs faster than it should. Or they are lower than they should be, deep and slow, and the speech drags. The picture is normal. The sound is not missing, not crackling, and not dropping in and out. It is present, and it is wrong in two ways at once: the speed and the pitch are both off, by the same amount, from the first second to the last.

Three details define the shape, and all three have to be present. First, the sound exists. A finished file with no sound at all belongs with the guide numbered 503. Second, the speed and the pitch move together, and they move by the same amount all the way through the file — the start is wrong by the same amount as the end. Third, the timeline itself played correctly before the export, or at least the source clips played correctly on their own; the wrongness appears in the exported file. Without that before-and-after, there is no baseline, and without a baseline the diagnosis has no evidence.

That combination is what makes this symptom different from its neighbours. The nearest neighbour is the guide numbered 496, which covers a recording file that plays smoothly in a player but stutters and drifts inside an editing program. In that shape the picture and the sound drift further and further apart towards the end of the clip, while the pitch itself stays normal. This guide starts from the opposite observation: the pitch is wrong, and the wrongness is the same at the start and at the end. Another neighbouring shape is a finished edit with sound but no picture — that is covered in the guide numbered 500, and its shape is a missing picture, not a wrong sound. A third neighbour is a finished edit whose preview colours are right but whose exported colours are wrong — that is covered in the guide numbered 499, and it is a colour label problem, not a sound sample rate problem. A fourth neighbour is sound that crackles and breaks up only while a game is running — that is covered in the guide numbered 311, and its shape is degraded sound, not changed pitch. A fifth neighbour is a television where the lips and the sound do not line up by a fixed amount, with the pitch normal — that is covered in the guide numbered 317, whose opening section states plainly that the sound has no crackling, no breaking up, and no pitch change, and pushes pitch change out of its scope. This guide is the topic that section pushes away.

The most important sentence in this guide has to be stated before anything else, and it comes from a guide on studiorescue.net: speed and pitch always move together, and that is the giveaway. If the timing is wrong but the pitch is right, this is not the problem this guide covers. A wrong sample rate changes both at once, because one wrong number decides both how fast the samples are played and how high they sound. If only the timing is wrong and the pitch is correct, the fault is somewhere else — most likely the variable frame rate road in the guide numbered 496, or a fixed offset of the kind in the guide numbered 317.

Two honesty points have to be stated alongside it. First, most published sources describe sample rate mismatch at the import or recording stage — a clip imported into an editing program, or a take recorded at the wrong rate — while this guide is about the exported file. The mechanism is the same at a different stage: a label that says the wrong sample rate, against data recorded at the right one. A guide on umatechnology.org states the export version of this plainly: the sync looks fine in the timeline but breaks after export. This guide never writes "the export setting is definitely the one that is wrong". It writes that the mechanism is the same, and uses the three comparisons below to find which segment is wrong. Second, a sample rate mismatch has two faces, and both have to be described. Measured by pitch, it looks like a uniform change of speed and pitch. Measured against the picture, it looks like a drift that grows with the length of the clip. Both are the same mechanism seen from different sides: played at the wrong rate, the whole clip runs uniformly fast or slow, and against the picture the error accumulates with time. This guide uses "speed and pitch move together" as its test, and never writes only one face of the problem.

One practical point has to be settled before anything else. A baseline must be written down before any test is run. The baseline is the editing program and its version, the sequence sample rate, the finished file's format and codec, and the length of the clip, together with the answers to three questions: is the whole clip wrong, or only some parts? Is the speech faster or slower, and by roughly how much? Is the pitch higher or lower? Without that baseline, "it sounds fast" is a feeling, and a feeling cannot separate a label fault from a sequence fault. The baseline should be written down, not remembered afterwards. Everything in this guide builds on it.

How a Sample Rate Works: A Label, Not the Sound Itself

Sound becomes digital by being measured many times per second. The sample rate is how many times per second the sound is measured: 44,100 means 44,100 samples every second, and 48,000 means 48,000 samples every second. That explanation comes from a reply by DVDdoug on the Audacity forum, and it is the whole mechanism in one sentence. Inside the file, besides the sound data itself, there is a label field that records what the sample rate is. In the WAV format that field is called `nSamplesPerSec`, as quoted by studiorescue.net from the Microsoft format documentation. When a player plays the file, it reads that label and decides how many samples to send out every second.

If the file was actually recorded at 48,000 samples a second, but the label says 44,100, the player will send out only 44,100 samples a second. The sound comes out slower by about 8 percent, and the pitch drops with it by about 1.5 semitones. The other way around is the mirror image: recorded at 44,100, labelled 48,000, and the sound comes out faster by about 9 percent, with the pitch about 1.5 semitones higher. The sound data itself is completely undamaged. Only the number on the label is wrong. Studiorescue.net states this directly: a wrong sample rate does not damage anything; it is a labelling problem, and labelling problems are reversible.

That mechanism explains why a change of speed and a change of pitch always appear together: one wrong label decides both. It also explains why resampling and relabelling are two different things, and why they must never be confused. Resampling converts the sound data itself to a different sample rate, re-interpolating the data, and the speed and the pitch do not change. Relabelling changes only the number written on the label, the data does not move, and the speed and the pitch return to normal. When the data is right and the label is wrong, the label should be changed. When the data itself was genuinely recorded at the wrong rate, the data has to be resampled. Until it is known which one is wrong, neither method can be called correct.

There is a second mechanism to understand, and it belongs to clocks. When a recording is made with more than one device, each device can act as its own clock master, or an interface can be set to a clock of 48 kHz while the source is actually running at 44.1 kHz. A support document from Lynx Studio Technology states the cause plainly: mismatched clock rates, or multiple clock masters. The label written into the file and the rate actually used then disagree. Video work normally uses 48 kHz — the AES5 recommendation is 48 kHz for professional production, processing, and exchange — while music and CD-era software mostly defaults to 44.1 kHz. Two different defaults in the same workflow are the most common source of a mismatch: the music side of the chain assumes 44.1 kHz, the video side assumes 48 kHz, and somewhere between them the label and the data part company.

The numbers that identify this fault are worth writing down, because they are the fingerprint. A guide on audioutils.com gives the gap between 44.1 kHz and 48 kHz as about 8.7 percent, and a ten-minute clip accumulating about 52 seconds of error. Studiorescue.net gives the same pair in a comparison table: 48 kHz recorded and labelled 44.1 kHz comes out about 8 percent slow and 1.5 semitones low; 44.1 kHz recorded and labelled 48 kHz comes out about 9 percent fast and 1.5 semitones high. Those numbers are the sources' own numbers, quoted as the sources' numbers, not as measurements made for this guide. They matter because they turn a vague "it sounds fast" into a testable claim: if the speed difference is about 8 to 9 percent and the pitch difference is about 1.5 semitones, the 44.1 and 48 kHz pair is the fingerprint. If the speed difference is some other number — 4 percent, or 25 percent — it points at a different pair of rates, or a different layer altogether, and the 44.1 and 48 kHz pair must not be forced onto it.

The Deciding Forks: Three Comparisons That Narrow the Field

The whole guide turns on three forks. Each fork is a comparison, and each comparison removes a whole set of possibilities. They should be done in order, before any setting is changed, because a setting changed before a fork is tested destroys the evidence the fork would have given.

Fork One: Is the Error the Same at the Start and at the End?

The first fork is the uniformity comparison, and it is the cheapest test in the guide. Find an obvious point where sound and picture meet — a clap, a door closing — near the start of the clip and another near the end, and compare the two. If both ends are wrong by the same amount, with the pitch changed by the same amount at both ends, the fault is locked into the sample rate layer, which is what this guide follows. If the start is right and the end drifts further and further out of sync, while the pitch itself stays normal, the fault is the variable frame rate road in the guide numbered 496. If both ends are wrong by the same amount but the pitch is normal and the speed is not the problem, that is a fixed offset, and it belongs with the guide numbered 317. This fork is first because it separates this guide from the guide numbered 496 in one comparison, and it costs almost nothing.

Fork Two: Does the File's Sample Rate Match the Sequence's Sample Rate?

The second fork is the MediaInfo comparison. Open the finished file in MediaInfo, look at the audio track, and read its sampling rate. Then read the sample rate of the editing sequence or project. If the two do not match — most commonly 44.1 kHz against 48 kHz — the fault is locked into the sample rate layer. If the two match, move to the clock and export layers in step six. The MediaInfo result has to be written down, because the two numbers side by side are the evidence that separates a label fault from a guess.

Fork Three: Does Every Player Hear the Same Wrongness?

The third fork is the player comparison. Play the same finished file in VLC and in another player, or back inside the editing program. If it sounds wrong in both places, the fault is in the file's label layer — the file itself carries the wrong sample rate. If it sounds right in VLC but wrong in the editing program, the fault is in the sequence interpretation layer: the editing program is reading the file as the wrong sample rate. An Adobe Community case is exactly this shape: 44.1 kHz audio placed into a 48 kHz sequence, producing a high-pitched tone, while the original clip played perfectly normally in Windows Media Player. The result of this fork decides whether the investigation follows the file or the sequence, and the two roads lead to different next steps.

ForkComparisonResult AResult B
Uniformity comparisonCompare a sound-and-picture point near the start with one near the endSame error at both ends, pitch changed together — sample rate layer, follow this guideStart right, end drifting, pitch normal — variable frame rate road, follow the guide numbered 496
MediaInfo comparisonRead the finished file's audio sampling rate and compare it with the sequence sample rate44.1 kHz against 48 kHz, or the other way around — sample rate layerBoth the same — move to the clock and export layers
Player comparisonPlay the same finished file in VLC and in another player or the editing programWrong in both — file label layerRight in VLC, wrong in the editing program — sequence interpretation layer

The Five Suspicion Layers

The five layers below are the whole map of this symptom. They are ordered by the cost of testing them, not by how likely they are, and the order itself is part of the diagnosis. The uniformity comparison is first because it costs almost nothing and it separates this guide from the guide numbered 496. The MediaInfo comparison is second because its result decides whether the problem is in the file itself. The player comparison separates the file layer from the sequence layer, and the two sides lead to completely different next steps. The clock layer and the export layer are only worth suspecting after the first three layers have been excluded.

LayerWhat it isDecisive test
First layer: file labelThe file's label records the wrong sample rate — the interface ran at one rate and the file recorded another — while the sound data itself is undamagedMediaInfo comparison: the finished file's audio sample rate against the sequence or project sample rate; proportion measurement: is the speed difference about 8 to 9 percent?
Second layer: sequence or project settings44.1 kHz audio placed into a 48 kHz sequence, or the other way around, so the editing program reads it at the wrong rateSequence comparison: what sample rate the sequence is set to inside the editing program; a file that plays right in VLC but wrong in the editing program points at this layer
Third layer: player interpretationThe player sends out samples according to the label, and different players handle the same label differentlyPlayer comparison: play the same file in VLC and in another player, and see whether both sound wrong or only one does
Fourth layer: device clockMore than one clock master, or an interface clock that does not match the source sample rateClock comparison: whether the recording interface settings have only one clock master, and whether the interface sample rate matches the source sample rate
Fifth layer: export or transcodingThe timeline plays correctly but the exported file is wrong; the export resampled without relabelling, or the export settings do not match the sequenceExport comparison: export a short clip from the same timeline and read its sample rate in MediaInfo; a timeline that plays right and a finished file that is wrong points at this layer

First Layer: File Label

This is the layer the MediaInfo comparison tests. The label says one sample rate and the data was recorded at another, and the player obeys the label. The sound data is undamaged, which is why relabelling — changing the number on the label — can return the speed and the pitch to normal without touching the data. The fingerprint test for this layer is the proportion measurement in step six: a speed difference of about 8 to 9 percent, together with a pitch difference of about 1.5 semitones, matches the 44.1 and 48 kHz pair. A different proportion points elsewhere, and the pair must not be forced onto it.

Second Layer: Sequence or Project Settings

This is the layer the sequence comparison tests. A 44.1 kHz clip placed into a 48 kHz sequence is read at the wrong rate by the editing program, and the result is the high-pitched, too-fast sound of the Adobe Community case — a case in which the original clip played perfectly normally in Windows Media Player, and only went wrong after being imported. The test is to read the sequence sample rate inside the editing program and write it down next to the file's sample rate, rather than assuming they must match. The conclusion of that Adobe Community case is quoted here because it is the rule in one sentence: make sure the sequence and the footage have the same audio sample rate.

Third Layer: Player Interpretation

This is the layer the player comparison tests. A player is not broken for playing a wrongly labelled file wrongly — it is doing exactly what the label tells it to do, sending out samples at the rate the label records. If two players disagree about the same file, the difference is in how each one reads the label, and the file's own label is still the evidence that matters. Blaming the player is the mistake this layer exists to prevent: the player is the messenger, and the label is the message.

Fourth Layer: Device Clock

This is the layer the clock comparison tests. When a recording is made with more than one device, or an interface clock is set to 48 kHz while the source is actually running at 44.1 kHz, the label written into the file and the rate actually used will disagree. The Lynx Studio Technology support document states the cause as mismatched clock rates, or multiple clock masters. The test is to check that the recording interface settings have only one clock master, with every other device set as a subordinate, and that the chosen sample rate matches the audio source. This layer is only reached after the first three layers have been excluded, because a clock fault is harder to test than a label fault, and testing it first would waste the evidence the cheaper comparisons give.

Fifth Layer: Export or Transcoding

This is the layer the export comparison tests. The timeline plays correctly, but the exported file is wrong. A guide on umatechnology.org describes this shape directly: the sync looks fine in the timeline but breaks after export. The export may have resampled without relabelling, or the export settings may not match the sequence. The test is to export a short clip from the same timeline and read its sample rate in MediaInfo. If the short export carries the same wrong rate, the fault is in the export layer. If the short export is correct, the fault is elsewhere, and the investigation returns to the earlier layers. Note the honesty point again: the Adobe Community case went wrong at import, while this layer goes wrong at export. The stage is different and the mechanism is the same, and the two must not be written as if they were the same case.

The Seven Diagnostic Steps

Work through these steps in order. Each step is a comparison, and each comparison produces a written result. Change only one thing at a time. A step that changes two things at once produces a result that cannot be read, because no one can tell which change did the work.

StepWhat to doWhat the result means
Step 1: Record the baselineWrite down whether the whole clip is wrong or only some parts, whether the speech is faster or slower and by roughly how much, whether the pitch is higher or lower, and whether the timeline plays correctlyWithout a baseline, "it sounds fast" is a guess, and the rest of the diagnosis has no evidence
Step 2: Uniformity comparisonCompare a sound-and-picture point near the start with one near the endSame error at both ends with the pitch changed together means the sample rate layer; start right and end drifting with a normal pitch means the guide numbered 496
Step 3: MediaInfo comparisonRead the finished file's audio sampling rate and compare it with the sequence sample rate44.1 kHz against 48 kHz locks in the sample rate layer; both the same means move to step 6
Step 4: Player comparisonPlay the same finished file in VLC and in another playerWrong in both means the file label layer; right in VLC and wrong in the editing program means the sequence interpretation layer
Step 5: Sequence comparisonCheck the sequence sample rate inside the editing program44.1 kHz audio in a 48 kHz sequence is the Adobe Community shape; write both numbers down
Step 6: Proportion measurementMeasure the speed difference: finished length against expected lengthAbout 8 to 9 percent with about 1.5 semitones of pitch change is the 44.1 and 48 kHz fingerprint; another proportion points at another pair of rates or another layer
Step 7: Decide and stopWrite down every comparison result and decide which layer the evidence points atIf the seven steps cannot separate the layers, or the sound also has crackling, breaking up, or distortion, stop and bring the records for inspection

Step 1: Record the Baseline

The first step is to describe the fault precisely. Is the whole clip wrong from start to finish, or only some parts? Is the speech faster or slower, and by roughly how much? Is the pitch higher or lower? Is the picture itself normal? Does the timeline play correctly? Write down the baseline: the editing program and its version, the sequence sample rate, the finished file's format and codec, and the length of the clip. A number read from a measurement is evidence. A feeling remembered afterwards is not. This step takes a few minutes and it protects every step that follows, because a diagnosis built on a guessed baseline is a guess all the way down.

Step 2: The Uniformity Comparison — The Decisive Test

Find an obvious point where sound and picture meet — a clap, a door closing — near the start of the clip and another near the end, and compare the sound-and-picture error and the pitch at both ends. If both ends are wrong by the same amount and the pitch has changed together with the speed, the fault is locked into the sample rate layer, and the investigation moves to step three. If the start is right and the end drifts further and further out of sync while the pitch stays normal, follow the road in the guide numbered 496 instead. If both ends are wrong by the same amount but the pitch is normal and the speed is not the problem, that is a fixed offset, and it is not this guide either. This is the decisive test of the whole guide, and it has to be run cleanly: the same clip, the same two points, nothing else changed.

Step 3: The MediaInfo Comparison

Open the finished file in MediaInfo, look at the audio track, and read its sampling rate. Then read the sample rate of the editing sequence or project. If the two do not match — 44.1 kHz against 48 kHz, or the other way around — the fault is locked into the sample rate layer, and the investigation moves to step four. If the two match, move to step six, the clock and export layers. The two numbers have to be written down side by side. A feeling that the rates "should match" is not a measurement, and it is exactly the feeling that lets a mismatch survive unnoticed.

Step 4: The Player Comparison

Play the same finished file in VLC and in another player. If it sounds wrong in both, the fault is in the file's label layer. If it sounds right in VLC but wrong in the editing program, the fault is in the sequence interpretation layer. This step's result decides whether the investigation follows the file or the sequence, and the two roads lead to different next steps, which is why the comparison has to be run before either side is investigated. Do not present a player that plays the file correctly as proof that the file is correct — it is proof that the player reads the label differently, and the label is still the evidence.

Step 5: The Sequence Comparison

Inside the editing program, check the sequence sample rate — in Premiere, through the audio channels view or the sequence settings — and write it down next to the file's sample rate. A 44.1 kHz clip placed into a 48 kHz sequence is the Adobe Community shape, and it is worth naming because it is the most concrete published case of this layer: the original clip played perfectly in Windows Media Player, and only went wrong after being imported. Write both numbers down. Do not write "should match" from memory.

Step 6: The Proportion Measurement

Measure the speed difference: the finished clip's length against its expected length. A ten-minute clip that is about 52 seconds out, which is about 8.7 percent, is the number a guide on audioutils.com gives for the 44.1 and 48 kHz pair. Together with a pitch difference of about 1.5 semitones, that proportion is the fingerprint of the pair: 48 kHz recorded and labelled 44.1 kHz comes out about 8 percent slow and 1.5 semitones low, and 44.1 kHz recorded and labelled 48 kHz comes out about 9 percent fast and 1.5 semitones high, according to the comparison table on studiorescue.net. If the speed difference is another number — 4 percent, or 25 percent — it points at a different pair of rates, or at a different layer altogether, and the 44.1 and 48 kHz pair must not be forced onto it.

Step 7: Decide and Stop

Once a layer has been found, write down every comparison result: whether the error is uniform or drifting, the MediaInfo sample rate, the player comparison result, the sequence sample rate, and the measured speed difference. A written record is what makes the next step possible, whether that step is a settings comparison, a clock comparison, or an inspection. If all seven steps have been run and the layers still cannot be separated, or if the sound is not a pure pitch change — if there is also crackling, breaking up, or distortion — stop testing and bring the full record for inspection. And do not delete or overwrite the original finished file before the layers have been separated, and never treat a suspect file as the only copy: while the original timeline and source material are still there, a new export can produce sound that is normal again.

Six Misconceptions That Mislead This Diagnosis

1. "The pitch changed, so the recording itself must have changed."

Wrong. The sound data is usually undamaged, and the number on the label is what is wrong. Studiorescue.net states it plainly: if a take plays slow and deep, or fast and high-pitched, the recorded samples are almost certainly fine. Until MediaInfo has been read, the recording itself must not be declared broken.

2. "A pitch change means the file is damaged and cannot be saved."

Wrong. A labelling problem is reversible: change the number on the label back, and the speed and the pitch return to normal. A label error must not be written as file damage. The data is intact; only the label is wrong.

3. "Resampling or transcoding will definitely fix it."

Wrong. Resampling converts the data itself, and relabelling changes the label number — the two fix different errors. When the data is right and the label is wrong, resampling will not fix it; it will write the wrong speed permanently into a new file. The data error and the label error have to be separated first, before either method can be called correct.

4. "44.1 kHz and 48 kHz are almost the same, so the difference does not matter."

Wrong. The two differ by about 8.7 percent, which on a ten-minute clip accumulates to about 52 seconds of error, according to a guide on audioutils.com, and the pitch differs by about 1.5 semitones — completely audible. Video work normally uses 48 kHz, while music software mostly defaults to 44.1 kHz, and the two different defaults are the source of the mismatch. A small-looking difference in the numbers is a large difference in the result.

5. "Sound out of sync must always be a variable frame rate problem."

Wrong. A sample rate mismatch can also put sound out of sync. The difference is in the pitch: with a sample rate error, the pitch changes together with the speed; with a variable frame rate error, the pitch stays normal and the drift grows worse towards the end of the clip. Until the start and the end have been compared, and the pitch has been compared, neither layer can be declared the cause.

6. "The player plays it wrong, so the player must be broken."

Wrong. A player only sends out samples at the rate the file's label records. If the label is wrong, the player playing it wrong is the normal reaction. The player must not be blamed for the fault; the file's own sample rate label has to be read. Two players disagreeing about the same file is evidence about how each one reads the label, not evidence that either player is broken.

Comparison Table: What Each Result Points At

ObservationMost likely layerNext test
Same error at start and end, pitch changed togetherSample rate layerMediaInfo comparison, step 3
Start right, end drifting further out of sync, pitch normalVariable frame rate layerFollow the guide numbered 496
MediaInfo shows 44.1 kHz against a 48 kHz sequence, or the reverseSample rate layerPlayer comparison, step 4
Wrong in every playerFile label layerSequence comparison, step 5
Right in VLC, wrong in the editing programSequence interpretation layerSequence comparison, step 5
Speed difference about 8 to 9 percent, pitch about 1.5 semitones44.1 and 48 kHz mismatchProportion measurement, step 6
Speed difference some other number, such as 4 percent or 25 percentAnother pair of rates, or another layerDo not force the 44.1 and 48 kHz pair; return to the clock and export layers
Timeline plays right, finished file is wrongExport or transcoding layerExport comparison: export a short clip and read its sample rate
Sound also crackles, breaks up, or distortsNot a pure pitch changeStop and bring the record for inspection

What the Sources Disagree About

The sources for this guide do not all say the same thing, and the disagreements have to be written out rather than settled silently.

Drift or pitch change? A guide on audioutils.com describes a sample rate mismatch as a growing drift, in a synchronisation framework. A guide on studiorescue.net describes it as a uniform change of speed and pitch, in a pitch framework. Both are right, because they are two faces of the same mechanism: played at the wrong rate, the whole clip runs uniformly fast or slow, which is the pitch face; and against the picture, the error accumulates with the length of the clip, which is the drift face. This guide uses "speed and pitch move together" as its test, and never writes only one face of the problem.

Resample or relabel? A guide on audioutils.com says the repair is to find which device used the wrong sample rate and resample to the correct rate. A guide on studiorescue.net says the repair is to change the playback speed by the proportion, which is re-reading the label. The two fix different errors: data genuinely recorded at the wrong rate needs resampling, while data that is right with a wrong label needs relabelling or re-reading. Until it is known which one is wrong, neither method can be called correct.

Import or export? The Adobe Community case went wrong at import — a sequence and a file with different sample rates. A guide on umatechnology.org says the timeline plays correctly but the export is wrong, which is an export settings conflict. Both are sample rate mismatches, but the fault is at a different stage. This guide states the stage difference plainly, and never writes an import case as if it were an export case.

Metallic distortion or pure pitch change? The Adobe Community case had high-frequency metallic distortion as well as a pitch change, while the studiorescue.net case is a pure change of speed and pitch. The distortion may be an extra problem introduced when the sequence resampled the audio, not a necessary symptom of a sample rate mismatch itself. This guide never writes distortion as a required symptom of a sample rate error.

When to Stop Testing

Stop testing and bring the records for inspection if any of the following apply. All seven steps have been run and the layers still cannot be separated. The sound is not a pure pitch change — there is also crackling, breaking up, or distortion. The export fails partway through, throws an error, or stops before it finishes, which belongs with the guide numbered 476. The finished file has broken pictures, coloured blocks, or flashing frames, which belongs with the guide numbered 493. The picture is missing entirely while the sound is normal, which belongs with the guide numbered 500. In every case, the written record is what makes the inspection useful: whether the error is uniform or drifting, the MediaInfo sample rate, the player comparison result, the sequence sample rate, and the measured speed difference, all with the editing program, the sequence settings, and the clip length attached.

What to Bring for Inspection

Bring the computer, or at least bring the written record, and bring the numbers rather than the conclusions. The editing program and its version, the sequence sample rate, and the finished file's format and codec. The MediaInfo result — the audio track's sampling rate, and how it compares with the sequence sample rate. The result of the uniformity comparison, the player comparison, the sequence comparison, and the proportion measurement, each with the exact step that was run. The published three-tier inspection and repair pricing for this service is HK$400–2,800, and the exact tier depends on what the inspection finds; no price is quoted before the inspection, because the fault has not been located yet. The shop is at Room 1209, Dali Plaza, Shek Mun. The telephone number is 2186 8010, and the WhatsApp number is 6558 6806. There is no shop in Mong Kok, and there is no on-site service.

Frequently Asked Questions

1. My exported video sounds fast and high-pitched — or slow and deep. Is the recording damaged?

Probably not. A fast, high-pitched export with the picture normal points at a sample rate label, not at damaged sound data. The most common pair is 44.1 kHz recorded and labelled 48 kHz, which comes out about 9 percent fast and about 1.5 semitones high, according to the comparison table on studiorescue.net. The mirror image is the same mechanism in the opposite direction: 48 kHz recorded and labelled 44.1 kHz comes out about 8 percent slow and about 1.5 semitones low. The recorded samples are almost certainly fine; only the number on the label is wrong. Read the finished file in MediaInfo, compare its sampling rate with the sequence sample rate, and measure the speed difference before deciding anything.

2. How do I tell a sample rate problem from a variable frame rate problem?

Compare the start of the clip with the end. A sample rate problem is uniform from start to finish — the same speed error and the same pitch error at both ends, because speed and pitch always move together in a sample rate fault. A drift that grows worse towards the end of the clip, with the pitch sounding normal, points at the variable frame rate road in the guide numbered 496. Run the uniformity comparison first: compare a sound-and-picture point near the start with one near the end, and compare the pitch at both ends, before deciding which road to follow.

3. How big is the gap between 44.1 kHz and 48 kHz?

By the proportion. The gap between 44.1 kHz and 48 kHz is about 8.7 percent, and a ten-minute clip accumulates about 52 seconds of error, according to a guide on audioutils.com. Together with a pitch difference of about 1.5 semitones, that proportion is the fingerprint of the pair. If the speed difference is another number — 4 percent, or 25 percent — it points at a different pair of rates, or at a different layer altogether. Measure the finished clip's length against its expected length, write the number down, and do not force the 44.1 and 48 kHz pair onto a different proportion.

4. Should I resample the audio to fix it?

Not until it is known whether the data or the label is wrong. Resampling converts the sound data itself to a different sample rate, and relabelling changes only the number written on the label. When the data is right and the label is wrong, resampling will not fix it — it will write the wrong speed permanently into a new file. When the data itself was genuinely recorded at the wrong rate, the data has to be resampled. Run the MediaInfo comparison and the proportion measurement first, and do not present either method as a universal fix.

5. The file sounds right in VLC but wrong in my editing program. Which one is lying?

Neither. The file's label says one sample rate, and the two programs read it differently: VLC plays it at the labelled rate, while the editing program reads it through the sequence settings. A file that plays right in VLC but wrong in the editing program points at the sequence interpretation layer — most commonly 44.1 kHz audio placed into a 48 kHz sequence, which is the Adobe Community case. Read the sequence sample rate inside the editing program and write it down next to the file's sample rate.

6. The timeline plays fine, but the exported file is wrong. How is that possible?

Because the fault can sit at the export stage rather than in the timeline. A guide on umatechnology.org describes this shape directly: the sync looks fine in the timeline but breaks after export. The export may have resampled without relabelling, or the export settings may not match the sequence. The test is to export a short clip from the same timeline and read its sample rate in MediaInfo. A timeline that plays right and a finished file that is wrong points at the export layer, and the stage has to be named honestly: most published sources describe the mismatch at import or recording, and this guide applies the same mechanism at the export stage.

7. The sound is out of sync — is it always a variable frame rate problem?

No. A sample rate mismatch can also leave the sound out of sync, and the pitch is what tells them apart: with a sample rate fault the pitch moves with the speed, while with a variable frame rate fault the pitch stays normal and the drift grows worse towards the end. Do not call it a variable frame rate problem before comparing the start with the end and checking the pitch. The two faults have completely different causes, and following the wrong road will not fix the clip.

8. My recording file has no sound at all. How is that different from this guide?

Completely silent is a different shape, covered by the guide numbered 503: the sound was never written into the file, and the fault sits in the recording source, the Windows audio device, the driver audio component, or the track mapping. This guide is about sound that was written in but comes out at the wrong speed and pitch. The boundary between the two is the MediaInfo step: no audio track follows guide 503, and an audio track with the wrong pitch follows this one.

9. The player plays it wrong — is the player broken?

Probably not. A player only plays back at the rate written on the file's label, so a wrong label makes the player play it wrong as a normal reaction. If two players both play it wrong, the fault sits in the file's label layer; only when one player plays it right and another plays it wrong is the reading layer worth suspecting. Do not blame the player before reading the file's sample rate label in MediaInfo.

10. The sound is wrong and it also crackles and distorts. What now?

Then the fault is not a pure pitch change, and this guide's seven steps have done their job by proving that. A sample rate mismatch on its own changes speed and pitch together; crackling, breaking up, and distortion point at an additional problem, such as the interrupt latency and audio buffer road in the guide numbered 311, or an extra problem introduced when the sequence resampled the audio. Stop testing and bring the full record for inspection: the uniformity result, the MediaInfo sample rate, the player comparison result, the sequence sample rate, and the measured speed difference. The published three-tier inspection and repair pricing is HK$400–2,800, depending on what the inspection finds. The shop is at Room 1209, Dali Plaza, Shek Mun, telephone 2186 8010, WhatsApp 6558 6806.

Sources

Related articles

Related services

If you need Chip-level GPU repair, see these service pages: