Skip to content

feat(cs): add missing Czech user phrases and SYSTEM entries - #188

Open
lukaskaplan wants to merge 1 commit into
EdgeTX:mainfrom
lukaskaplan:cs-cz-add-missing-phrases
Open

lukaskaplan wants to merge 1 commit into
EdgeTX:mainfrom
lukaskaplan:cs-cz-add-missing-phrases

Conversation

@lukaskaplan

Copy link
Copy Markdown

Summary

  • voices/cs-CZ.csv only had 289 of the 649 entries available in
    en-US.csv, leaving 356 user-assignable phrases (Special/Global
    Functions "Play track") completely untranslated and 14 SYSTEM
    entries (decimal digits, dBm, km) missing.
  • Adds all 360 missing entries with standard English filenames
    (matching en-US.csv), so existing guides/scripts/shared models that
    expect standard names keep working — the previous 33 Czech user
    phrases used custom names (e.g. podvo0.wav instead of gearup.wav).
  • Generated with cs-CZ-VlastaNeural, the voice already configured for
    this locale in voice_generation_config.py, at 32kHz/16-bit/mono PCM
    as required by EdgeTX.
  • Abbreviations spelled out phonetically for correct TTS pronunciation
    (VTX -> "vé té iks", OSD -> "ó es dé", GPS -> "gé pé es", LQ -> "el kvé",
    BEC -> "bé é cé", etc).
  • Covers the Czech (cz) translations requested in Finish translating additional words/phrases added #17
    (manmod/acrmod/angmod/hznmod/lncmod/airmod/pitmod/navcrs/navcsh/navalh/
    navpos/navrth/navwp/cruise/crshld/althld/poshld/rth/waypnt/manual/auto/
    osd/osdprf) plus the rest of the missing catalogue.

Related: #17

Test plan

  • python build_checks.py passes with no errors
  • Spot-checked generated WAV files: 16-bit PCM, mono, 32000 Hz (ffprobe)
  • No duplicate String IDs/filenames; IDs 290-649 sequential, no gaps
  • pytest not run locally (module not installed here) — no Python
    code was changed, only voices/cs-CZ.csv and SOUNDS/cz/*.wav

🤖 Generated with Claude Code

https://claude.ai/code/session_01HT2RZGNHz58E73Ng32R7ru

The cs-CZ voice pack only covered 289 of the 649 phrases available in
en-US, leaving 356 user-assignable phrases (Special/Global Functions
"Play track") completely untranslated and 14 SYSTEM entries (decimal
digits, dBm, km) missing.

Adds all 360 missing entries to voices/cs-CZ.csv with standard English
filenames (matching en-US, so existing guides/scripts/shared models
keep working) and generates the corresponding WAV files using the
cs-CZ-VlastaNeural voice already configured for this locale in
voice_generation_config.py, at 32kHz/16-bit/mono as required by EdgeTX.

Abbreviations are spelled out phonetically for correct TTS pronunciation
(VTX -> vé té iks, OSD -> ó es dé, GPS -> gé pé es, etc.).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HT2RZGNHz58E73Ng32R7ru
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant