Ixil Speech to Text: Preserving a Mayan Language with AI

Ixil is a Mayan language spoken by around 100,000 people in the highlands of Guatemala, primarily in the departments of Quiché and Huehuetenango. It belongs to the Eastern Mayan branch and has three main dialects: Nebaj, Chajul, and Cotzal. The language is rich in oral tradition, but written resources are scarce. For many Ixil speakers, Spanish is the language of literacy, and Ixil remains primarily spoken. As a result, there is a growing need for tools that can convert spoken Ixil into written text, both for documentation and for practical communication in the digital age.

Why Accurate Speech-to-Text Matters for Ixil

Automatic speech recognition (ASR) for Ixil can serve multiple purposes. Community radios, which broadcast in Ixil, can quickly generate transcripts for archives or for translation into Spanish. Linguists studying the language can process hours of fieldwork recordings efficiently. Educators can create subtitles for Ixil-language teaching materials, making them accessible to learners who read the language. Moreover, speech-to-text can help Ixil speakers access voice-based technology — from dictation on smartphones to voice search — in their own language.

The Linguistic Challenges of Transcribing Ixil

Ixil presents several challenges for ASR systems:

• Ejective and glottalized consonants: Ixil has a series of ejective stops (p', t', k', ch', q') and glottalized sonorants (m', n', l', y', w'). These sounds are rare in world languages and require careful acoustic modeling.
• Pitch accent: Words can be distinguished by a high or low pitch accent. For example, 'wal' (thing) vs. 'wál' (to sleep) differ only in pitch. The model must capture this suprasegmental feature.
• Dialectal variation: The three main dialects differ in pronunciation, vocabulary, and even some grammatical features. A model trained on Nebaj may not perform well on Cotzal speech.

Speechyou addresses these challenges by providing a dedicated Ixil model trained on a balanced corpus of all three dialects. Users can select their dialect before transcription, and the system adapts its recognition accordingly. Custom vocabulary uploads allow users to add local place names or personal names.

Use Cases in Action

• Preserving oral histories: Indigenous organizations can transcribe interviews with elders, creating digital archives of stories and knowledge.
• Subtitling community videos: A YouTube channel in Ixil can generate SRT subtitles automatically, reaching a wider audience that includes Spanish speakers.
• Language learning: Students of Ixil can practice by reading transcripts of native speech. The combination of audio and text reinforces vocabulary and pronunciation.
• Accessibility: Deaf or hard-of-hearing Ixil speakers can read captions for video content, though the primary audience is hearing individuals who need text.

How Speechyou Helps

Speechyou is the only AI transcription platform that supports Ixil with native accuracy. It handles audio files, YouTube links, and direct recording. The output can be exported as plain text, SRT, or VTT subtitles. For institutions working with limited budgets, the Solo plan includes unlimited transcription, making it affordable to process large volumes of Ixil audio. The platform also supports over 100 other languages, so users can transcribe multilingual content within the same project.

To get started, simply upload an Ixil audio file, select the dialect, and receive a transcript in minutes. The interface is intuitive and requires no technical expertise. With Speechyou, the Ixil language gains a powerful tool for digital preservation and communication.