Check the archive you actually saved
A completed preparation is not the same as a completed download. First check the browser's download list, locate the ZIP and extract it into a new folder. Opening a clean folder makes it easier to distinguish this export from files saved during an earlier attempt. Confirm that the archive contains the audio you intended to keep, together with tracks.csv and manifest.json.
You can export a CSV without building a ZIP. That standalone CSV describes the collection at the time you export it. If you then change the selected tracks or naming style, a later ZIP can legitimately differ. For a precise comparison, use the CSV included inside the same ZIP you are checking. See the batch workflow for preparing and rebuilding an archive.
Read the columns as a record of preparation
Open the CSV in a spreadsheet application, using UTF-8 when the import dialog asks for an encoding. The file uses quoted fields and includes a UTF-8 marker to help with non-English titles. Keep a copy of the original CSV before editing it. A spreadsheet may reformat long identifiers, numbers or text when saving, so an edited workbook is not automatically an exact copy of the original record.
| CSV field | What to check |
|---|---|
title, song_id, source_url | Identify the intended song and distinguish similarly named takes. |
status, note | Read the preparation result and any reason an entry was not ready. |
included_in_zip | Use yes/no to check the selection represented by this export. |
filename | Match the named output to a file in the extracted folder. |
duration_seconds, duration | Compare measured seconds with the rounded, readable display. |
bytes, sha256 | Check the prepared file's size and, when needed, its integrity. |
The JSON manifest stores related information with names such as included, link and id. Those are JSON keys, not alternative CSV column headings. It also contains creation time, naming style and totals. Use CSV for a convenient table and JSON when you need the fuller structured record.
Why rows can outnumber audio files
The list includes retained inputs that failed, were cancelled, resolved as duplicates or were left unselected. Only selected ready files belong in the ZIP. For example, imagine three different inputs: one ready and selected, one ready but unchecked, and one failed. The record has three rows, but the ZIP should contain one audio file. This is an illustrative case, not a report of your current collection.
Two identical pasted URLs are merged before preparation, so they do not create two manifest rows. Two different links that resolve to the same song can leave a duplicate result in the list. This distinction explains why the number of pasted lines, visible results, CSV rows and saved audio files can all differ without data being lost. The duplicate and naming guide explains how source identity differs from a repeated title.
Filter the CSV to rows with a ready status and included_in_zip set to yes. Compare those filenames with the extracted audio files. Then inspect the excluded and failed rows separately. A missing audio file is expected for a failed row; it is a problem worth investigating when a selected ready row has no matching file in that same archive.
Compare lengths and sizes carefully
The numeric duration_seconds value can retain fractions of a second. The readable duration value rounds for display. If you add already rounded individual values, the result can differ slightly from a total calculated from the underlying seconds and rounded once. Use the numeric field when doing a precise spreadsheet calculation, and apply the same selection filter as the ZIP.
For a simple example, a selected 3:10 track and a selected 4:20 track total 7:30. Unselecting the second leaves a selected duration of 3:10 without removing it from the ready collection. Treat a large mismatch as a reason to inspect the files and selection. Small rounding differences alone do not show that audio is missing.
The bytes value is an exact prepared-file size, while a file manager may show a rounded MB or MiB figure. Compare actual bytes when possible. The ZIP archive size also includes records and archive overhead, so it is not a direct substitute for the sum of individual audio sizes. Open the files in your destination player as part of the check.
Use hashes for file identity, not permission
A SHA-256 value can help compare a saved audio file with the prepared bytes recorded in the manifest. If the file is edited, tagged or transcoded afterwards, its hash can change even when it still sounds similar. A matching hash indicates matching bytes; it does not establish the source's ownership, the song's quality, a licence or an authorised platform download.
Song titles are untrusted input. The CSV exporter quotes fields and prefixes formula-like text to reduce the chance that a spreadsheet interprets a title as a command. If you notice an initial apostrophe in such a cell, it may be that protective representation. Do not enable spreadsheet actions or external links merely because a source title suggests doing so.
Finally, review the records before sharing. Source URLs, IDs and titles can disclose the origin of a collection. Keep rights records separately and check download and sharing permissions. If the archive differs from its own manifest, save both for your investigation, avoid posting private information publicly, and report the reproducible steps with the browser version and visible error.