333 Downloads view our datasets

Alp Öktem

  • LinkedIn

About Alp Öktem

7,000+ languages exist. AI works for maybe 100 of them. I’m working to fix that.

I build speech and language tech for languages the industry ignores—from Tamazight to Judeo-Spanish, Catalan to Hausa, because every language deserves a voice in the AI age.

PhD in computational linguistics, 10 years of working with messy data, reluctant communities, and zero-resource scenarios. I’ve built the first neural voice for Catalan, shipped production MT, ASR and TTS for organizations like UNICEF and Translators without Borders, and published at Interspeech, EMNLP, and beyond. My work has touched 30+ languages from Africa, Europe and Asia, including endangered ones.

What I do: turn “impossible” into working prototypes. ASR for impaired speech. MT for endangered languages. I work with NGOs, researchers, and language communities who need someone technical but also gets why your grandmother’s language matters.

Datasets

Alp Öktem

Araina Text Corpus (Occitan Aranese)

Text corpus in Aranese variety of Gascon dialect of Occitan
License Icon

License: CC0-1.0

Locale Icon

Locale: oc

Task Icon
Task: LM
Format Icon

Format: txt

Size Icon

Size: 22.97 MB

Alp Öktem

Awal Tamazight Dataset

A compilation of monolingual and parallel Tamazight language datasets created by CIEMEN as part of the Awal community language technology project.
License Icon

License: CC-BY-4.0

Locale Icon

Locale: zgh

Task Icon
Task: LM
Format Icon

Format: TSV, JSON, TXT

Size Icon

Size: 11.57 MB

Alp Öktem

Heroes English-Spanish Dubbed Movie Speech Corpus

7000 single speaker speech segments from the original and Spanish dubbed version of 21 episodes of TV series Heroes
License Icon

License: CC-BY-SA-4.0

Locale Icon

Locale: eng, spa

Task Icon
Task: NLP
Format Icon

Format: wav, csv, txt

Size Icon

Size: 1.68 GB

Alp Öktem

Ladino-Spanish Lexical Resources

Dictionary, word list, and verb conjugation files for Ladino (Judeo-Spanish) and Spanish, compiled for a rule-based translation system.
License Icon

License: CC-BY-4.0

Locale Icon

Locale: lad, spa

Task Icon
Task: MT
Format Icon

Format: TXT

Size Icon

Size: 39.92 KB

Alp Öktem

Ladino TTS Corpus

A single-speaker Ladino (Judeo-Spanish) speech corpus of ~2,000 aligned audio segments for training text-to-speech models.
License Icon

License: CC-BY-NC-4.0

Locale Icon

Locale: lad

Task Icon
Task: TTS
Format Icon

Format: WAV, TSV

Size Icon

Size: 196.00 MB

Alp Öktem

Ladino: Una Fraza al Diya

307 Ladino (Judeo-Spanish) language learning sentences with translations in Turkish, English, and Spanish, plus audio recordings and images.
License Icon

License: CC-BY-4.0

Locale Icon

Locale: lad

Task Icon
Task: NLP
Format Icon

Format: OGG, JPEG, TSV

Size Icon

Size: 76.35 MB

Alp Öktem

LezizNet - Turkish food images

3,272 openly-licensed (Creative Commons) images of Turkish food across 245 dish labels, with documented per-image source, creator, and license.
License Icon

License: CC-BY-SA-4.0

Locale Icon

Locale: tur

Task Icon
Task: CV
Format Icon

Format: JPG, TSV

Size Icon

Size: 784.29 MB

Alp Öktem

Şalom Ladino Corpus

Text corpus of 176,843 words compiled from 397 Judeo-Espanyol articles from Şalom newspaper.
License Icon

License: CC-BY-4.0

Locale Icon

Locale: lad

Task Icon
Task: LM
Format Icon

Format: TXT

Size Icon

Size: 403.16 KB

Alp Öktem

SKAD Ladino Parallel Corpus

Human-authored parallel sentences for Ladino (Judeo-Spanish) paired with English and Turkish, created by the Sephardic Center of Istanbul (SKAD).
License Icon

License: CC-BY-4.0

Locale Icon

Locale: lad, eng, tur

Task Icon
Task: MT
Format Icon

Format: TSV

Size Icon

Size: 74.14 KB

Alp Öktem

SKAD Ladino Speech Database (subset)

Spoken Ladino (Judeo-Spanish) from native speakers in Istanbul, recorded and transcribed by SKAD, re-packaged for ASR training and evaluation.
License Icon

License: CC-BY-NC-SA-4.0

Locale Icon

Locale: lad

Task Icon
Task: ASR
Format Icon

Format: WAV, TXT

Size Icon

Size: 4.99 GB

Alp Öktem

Synthetic Ladino Parallel Corpus

Synthetically produced parallel sentences for Ladino paired with English, Spanish, and Turkish, generated using rule-based translation.
License Icon

License: CC-BY-4.0

Locale Icon

Locale: lad, en, es, tr

Task Icon
Task: MT
Format Icon

Format: TSV

Size Icon

Size: 898.32 MB