Transcription serviceJapanese
Japanese Transcription Services
Japanese transcription with a written script policy across kanji, hiragana and katakana, backchannel conventions agreed in advance, and regional audio staffed by region.
Free sample: Send a short Japanese audio clip and we return a transcript in the format and orthography you specify, at no cost, before you commit to a project.
Specification
- Audio
- Wideband, narrowband, mono, or separated channels
- Script
- Kanji conversion policy for words writable in either kanji or kana, plus loanword handling, fixed up front
- Backchannels
- Aizuchi transcribed in full, sampled, or omitted, stated in the scope
- Accuracy
- Target agreed per project and measured on a held-out slice of your own recordings
- Review
- Second-pass review on a sampled percentage, rate agreed up front
Dialects and varieties covered
Tokyo, Kansai, Tohoku, Hakata.
Who does the work
- Contributor floor
- 1,000+per market
- Markets covered
- Japan
- Dialects covered
- 4dialects
Coverage is drawn from Japan, where Spirelight already publishes a delivered Japanese dataset, so it is demonstrated rather than asserted.
Three scripts, one utterance, and no space between the words
Every Japanese sentence is written in a mix of kanji, hiragana and katakana, and that mix is a choice rather than a property of the speech. The same spoken word can be written in characters or in kana, and two transcribers will draw the line in different places unless the line is drawn for them. Loanwords raise a second question: katakana as pronounced, or the Latin original as written.
Spirelight writes the conversion policy into the style guide before work starts, covering numerals, dates and company names as well. Without it, batches stay individually correct and collectively inconsistent, which is the version of the problem that only surfaces once you have trained on it.
Call audio is mostly overlap and acknowledgement
Japanese conversation is dense with aizuchi, the short acknowledgements a listener produces continuously while the other person is still speaking. In customer support recordings they can outnumber content words. Transcribing all of them, some of them, or none of them changes what a diarization or turn-taking model learns, so it is a decision to make deliberately rather than by default.
Keigo is the other register question. Honorific and humble forms are constant in service speech and encode who is speaking to whom, so verbatim delivery keeps them intact. Kansai, Hakata and Tohoku audio is assigned to annotators from those areas, and Okinawan material is scoped separately because it is not Japanese with an accent.
This service is not the right fit when
- Romaji-only transcripts. Delivery is in Japanese script unless a romanized field is specified as an extra layer.
- Japanese to English translation or timed subtitle files.
- Machine-only transcription with no human pass.
Related
Need Japanese transcription scoped to your project? Send the audio conditions, volume, and turnaround you need. We will reply with a scoped quote and, if it helps, a free sample.