# 调调 accented titles: separate two Café representations from recording identity

Language: en
Canonical: [调调 accented titles: separate two Café representations from recording identity](<https://allluckly.cn/en/guides/diaodiao-accented-title-nfc-nfd-codepoints-audio-identity/>)
中文：[中文 HTML](<https://allluckly.cn/guides/diaodiao-accented-title-nfc-nfd-codepoints-audio-identity/>)
English: [English HTML](<https://allluckly.cn/en/guides/diaodiao-accented-title-nfc-nfd-codepoints-audio-identity/>)
Published: 2026-11-19T00:00:00+08:00
Modified: 2026-11-19T09:38:09+08:00
Author: Bison

Prepare visually similar accented titles for two distinct owned recordings, retaining external NFC/NFD evidence before checking displayed titles and sound in 调调. Canonical equivalence is neither identical original code points nor identical audio. No preserved normalization form or in-app code-point inspector is promised.

English translation of the [Chinese original](<https://allluckly.cn/guides/diaodiao-accented-title-nfc-nfd-codepoints-audio-identity/>), translated on 2026-11-19. Original publication dates, attribution, code and version notes are retained.

Prepare visually similar accented titles for two distinct owned recordings, retaining external NFC/NFD evidence before checking displayed titles and sound in 调调. Canonical equivalence is neither identical original code points nor identical audio. No preserved normalization form or in-app code-point inspector is promised.

## Establish different audible recording identities

Prepare genuinely different owned recordings speaking 字形甲 and 字形乙, named accent-a.wav and accent-b.wav externally. Retitling identical audio would confound identity and deduplication. Establish readability and preserve originals/tagging project, fixing theme/output/speed. This examines title representation, not accented filenames. Some filesystems/tools normalize names; distinct ASCII paths avoid confusing path collisions with metadata. Unprepared recordings remain a target ledger, not supplied downloads.

## Preserve the original text of both title forms

Both targets read Café du port. A uses precomposed é; B uses e followed by a combining accent. Preserve both original strings with explicit code-point evidence externally rather than screenshots alone. A’s é is U+00E9; B uses U+0065/U+0301, with other letters/spaces identical. A separate combining code point is not automatically another standalone displayed letter. Code-point count is not a universal glyph count. A visually matching retyped title does not establish both forms; inspect the actual strings.

## Explain NFC/NFD equivalence without replacing originals

Unicode normalization distinguishes composed and decomposed forms. A is written in NFC and B in NFD; NFC processing can produce the same text without making the original strings code-point-identical. The complete title has 12 code points in A and 13 in B, not a claimed screen width/glyph count. Keep UTF-8 byte counts separate if inspected; extra bytes do not lengthen audio. Retain originals before normalized copies rather than normalize both and claim observation of untouched NFD input.

```
A (NFC): Café du port
B (NFD): Café du port
A code-point count=12; B code-point count=13
accent in A: U+00E9; accent in B: U+0065 U+0301
NFC(A)=NFC(B); original A != original B by code-point sequence
Different owned recordings remain different audio even when titles are canonically equivalent.
```

## Read back actual tags immediately after writing

Using an existing external tag editor, assign A/B targets to the different recordings while retaining artist/other conditions. Read actual title text or a reliable report afterward to see whether the tool preserved the input form. Seeing a combining accent in an entry field proves no stored code points. If normalized automatically, record readback; audio identity remains comparable but both original representations were not established in files. Without reliable inspection leave preservation pending. Invent no in-app tagging/inspection ability, and substitute no artwork for textual evidence.

## Associate imported title display with source sounds

Handle both files through actual import controls, recording added/deduplicated/failed outcomes before selecting identifiable sources. Similar titles require spoken A/B verification rather than an assumption of identical audio. Preserve path, readback title and actual sound without relying on permanent list order. Without two distinguishable sources, mark the display comparison limited instead of selecting one twice as proof of two encodings. Player titles follow actual read metadata; filenames remain separate external locators.

## Audit visual similarity separately from raw equality

Observe complete titles under one theme/layout, recording readable accents, missing glyphs or truncation. Do not change fonts and attribute every difference to normalization. The same visible Café supports readable display, not internal NFC/NFD/search semantics. If an owned compatible server offers an actual search entry, two original queries may form separately recorded extra conditions; this is no promised local-list feature. Keep query, returned title and recording identity separate from raw-string equality.

## Trace originals, tag readback and selected entries

For no difference ask whether the question is display or code points, then inspect earlier normalization. Identical sound calls for checking selected source/accidental copied recording. A missing accent needs original/tag inspection; font changes alone restore no character. For truncation use an actually available fuller title view without inventing a hidden details screen. Change one work-copy factor while retaining originals/results rather than remove all accents to pass. Keep internal ordering/deduplication/search rules unknown; one title pair proves no universal multilingual normalization support.

## Deliver original, file-readback and player ledgers

Retain A/B originals and code-point evidence, actual file readback and player display/sound, locating any normalization step. Explain why 12/13 code points can look alike and identical-looking titles can identify different recordings, without requiring disclosure of internal algorithms. Mark unperformed server queries/readback/playback individually pending and retain source/project. Restore normal appearance/queue; keep practice titles on work copies. New characters/fonts/formats/servers require new premises, not universal Unicode/tagging/device certification.

## Related guides and sources

[Metadata and source identity](<https://allluckly.cn/en/guides/diaodiao-metadata-title-artist-artwork-identity/>); [Long titles and audible identity](<https://allluckly.cn/en/guides/diaodiao-long-shared-prefix-titles-three-view-identity/>).

Checked on 19 November 2026: [Apple’s Russia lookup](<https://itunes.apple.com/lookup?id=6827171352&country=ru&entity=software&timestamp=20261119>) and [China lookup](<https://itunes.apple.com/lookup?id=6827171352&country=cn&entity=software&timestamp=20261119>) both list 调调 1.0.2, requiring iOS / iPadOS 17.6 or later. The [official store](<https://apps.apple.com/ru/app/id6827171352?l=en-GB>) describes audio import, playback, lyrics and music themes and Home Screen widgets. Use owned or authorized material and controls actually visible in the installed version. Keep external preparation, paper expectations and real observations separate; no completed device, private-server or exhaustive format-compatibility test is claimed.

Method references: [Unicode normalization forms](<https://www.unicode.org/reports/tr15/>).

