Chiapas Zoque Speech to Text: Preserving a Language with AI

Chiapas Zoque, known natively as O'de püt, is a language that carries the history and identity of around 70,000 people in the Mexican state of Chiapas. Part of the Mixe-Zoquean family, it is primarily spoken in municipalities such as Copainalá, Francisco León, Ostuacán, and Rayón. As a low-resource language, Zoque has limited digital presence, and most commercial speech recognition tools do not support it at all. This digital divide makes it difficult for speakers to access modern technologies like automatic transcription, subtitling, and voice interfaces.

Why Accurate Zoque Speech-to-Text Matters

For an indigenous language, speech-to-text is not merely a convenience; it is a tool for survival. Accurate transcription enables:

• Language documentation: Elders’ oral narratives can be turned into written texts for archives and school materials.
• Accessibility: Zoque-speaking individuals can access content via text where audio is not suitable.
• Preservation of endangered dialects: Each Zoque community has unique lexical and phonological features that deserve to be recorded faithfully.

Without dedicated ASR models, Zoque audio would remain locked in analog tapes or unsearchable digital files. Speechyou steps into this gap with a system trained specifically on Zoque phonetics and dialects.

Specific Challenges in Zoque Transcription

The Zoque language presents several hurdles for automatic recognition:

• Ejective consonants (pʼ, tʼ, kʼ, tsʼ, chʼ): These sounds involve a simultaneous closure of the glottis, producing a sharp burst. Most ASR models confuse them with their plain counterparts.
• Glottal stop /ʔ/: Frequently occurs between vowels and at syllable boundaries. It is often omitted by generic models.
• Tonal distinctions in some dialects: Copainalá Zoque uses pitch to differentiate words; e.g., kayu (house) vs. kàyu (turtle).
• Vowel length and nasalization: Contrastive in certain Zoque varieties and easily misheard by untrained systems.

Speechyou’s model handles these features through a phoneme set based on linguistic fieldwork and dialect-specific acoustic models.

Use Cases for Zoque Transcription

• Subtitling community videos: Add Zoque or Spanish subtitles to cultural videos using the generated SRT files.
• Research and lexicography: Transcribe fieldwork interviews quickly to build bilingual dictionaries.
• Legal and medical interpreting: Ensure accurate records when Zoque speakers interact with Spanish-speaking institutions.
• Education: Create transcribed reading materials from oral stories for Zoque-language literacy classes.
• Podcasting: Zoque podcasts, such as Nukujü Püt (Our Language), can be transcribed for show notes and searchability.
• Oral history preservation: Digitize and transcribe cassette recordings from the 1980s and 90s before they degrade.

How Speechyou Delivers Accurate Zoque Transcription

Rather than relying on a one-size-fits-all model, Speechyou offers:

• Dialect selection: Choose between Copainalá, Francisco León, or Ostuacán before uploading audio.
• Custom vocabulary: Add community-specific terms (place names, clan names, neologisms) to improve recognition.
• Speaker diarization: Distinguish up to four speakers in a Zoque conversation.
• Export to SRT/VTT: Directly generate subtitles for videos in Zoque.

Our ongoing collaboration with Zoque linguists and community members ensures that the model improves over time. We are committed to making speech technology inclusive for every language, no matter how many speakers it has.

Getting Started

Transcribing Zoque audio with Speechyou is straightforward:

1. Upload your audio or video file (MP3, WAV, MP4, etc.).
2. Select Chiapas Zoque as the language and choose your dialect.
3. Receive a full transcript with timestamps, editable online.
4. Export as SRT, VTT, or plain text.

Whether you are a linguist, a community activist, or a Zoque speaker wanting to preserve your family stories, Speechyou is here to turn your speech into text. Try it today and give O'de püt a voice in the digital world.