Transcription serviceSpanish
Spanish Transcription Services
Spanish transcription staffed by the variant in the recording, so Rioplatense audio never lands with an annotator trained on Madrid broadcast speech, and the output convention is fixed before the first file is processed.
Free sample: Send a short Spanish audio clip and we return a transcript in the format and orthography you specify, at no cost, before you commit to a project.
Specification
- Variant
- Staffed by market: Spain and Latin America are separate pools, never one Spanish queue
- Audio
- Wideband, narrowband, mono, or separated channels
- Accuracy
- Target agreed per project and measured on a held-out slice of your own recordings
- Output
- Verbatim or clean-read text, timestamps, speaker turns, voseo kept unless you ask otherwise
- Review
- Second pass by a reviewer from the same variant, sampled at an agreed rate
Dialects and varieties covered
Castilian, Andalusian, Mexican Spanish, Rioplatense, Bogotá Spanish, Costeño.
Who does the work
- Contributor floor
- 1,000+per market
- Markets covered
- 2markets
- Dialects covered
- 6dialects
Coverage is drawn from Spain and Latin America, where Spirelight already publishes delivered Spanish datasets, so it is demonstrated rather than asserted.
Castilian, Rioplatense and Mexican are not interchangeable
The grammar splits before the accent does. A Buenos Aires speaker says vos tenés where a Madrid speaker says tú tienes, and vosotros forms appear in Spain and almost nowhere else in volume. An annotator writing what the textbook predicts rather than what the speaker produced will quietly regularize voseo out of your data.
Lexis is the second split: coche, carro and auto name the same object in three markets, and diminutives, address terms and filler words move with them. Spirelight staffs Spanish by the variant present in the audio and keeps the speaker's form in the transcript, unless you ask for normalization and that decision goes into the style guide.
Aspirated endings are where the accuracy number is won or lost
Caribbean and Andalusian speakers aspirate or drop syllable-final s, so the plural marker and the second person ending are frequently not in the signal at all. The transcriber recovers them from agreement elsewhere in the phrase. That is a judgment call, and judgment calls are exactly what an error rate measured on clean studio audio hides.
Spirelight measures the target on a held-out slice of your own recordings, and the scope states what counts as an error when the ending is inaudible. The sampled review pass runs inside the same variant pool, because a reviewer from another market reintroduces their own defaults and calls it a correction.
This service is not the right fit when
- Catalan, Galician or Basque audio. Those are separate languages, not Spanish accents, and they are scoped and staffed as their own work.
- Machine-only transcription with no human pass. Spirelight does not resell raw ASR output.
- Translation between Spanish and English. Transcription stays in the language of the recording.
Related
Need Spanish 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.