Q'anjob'al Speech to Text: Preserving a Mayan Language with AI

Q'anjob'al (also spelled Kanjobal) is a Mayan language spoken primarily in the highlands of Huehuetenango, Guatemala, with around 100,000 native speakers. It is one of 21 Mayan languages recognized in Guatemala and has a rich oral tradition. However, like many indigenous languages, Q'anjob'al faces pressures from Spanish dominance, and digital resources for the language are scarce. Accurate speech-to-text technology can play a vital role in documentation, education, and media creation.

Where is Q'anjob'al Spoken?

Q'anjob'al is concentrated in municipalities such as San Juan Ixcoy, Santa Eulalia, San Miguel Acatán, and San Juan Cotzal. There are also significant diaspora communities in the United States, particularly in Florida and California. The language has three main dialect groups:

• Eastern Q'anjob'al: Spoken in San Juan Ixcoy and San Pedro Soloma.
• Western Q'anjob'al: Spoken in Santa Eulalia and San Miguel Acatán.
• Central Q'anjob'al: Spoken in San Juan Cotzal and Chajul.

Each dialect has its own phonetic and lexical characteristics, which must be accounted for in speech recognition.

Why Accurate Transcription Matters

For Q'anjob'al speakers, being able to transcribe audio to text opens up new possibilities:

• Language preservation: Record and transcribe elders' stories, traditional songs, and ceremonies.
• Education: Create bilingual teaching materials and subtitles for instructional videos.
• Media: Add Q'anjob'al subtitles to community news broadcasts and YouTube content.
• Accessibility: Provide captions for deaf or hard-of-hearing community members.
• Research: Help linguists and anthropologists analyze spoken Q'anjob'al.

Without reliable ASR, these tasks require manual transcription, which is time-consuming and expensive.

Challenges in Q'anjob'al Speech Recognition

Developing ASR for Q'anjob'al is not trivial. Key challenges include:

• Ejective consonants: Sounds like /p'/, /t'/, /k'/, /q'/, /ch'/ are common and must be distinguished from plain stops.
• Vowel length: Long and short vowels are contrastive, e.g., 'sat' (face) vs 'saat' (to look).
• Glottalization: Many consonants are glottalized, requiring fine acoustic discrimination.
• Dialectal differences: Pronunciation varies, so a one-size-fits-all model may fail.

Speechyou's AI has been trained on a diverse corpus of Q'anjob'al speech from multiple dialects, with special attention to these phonological features. The result is a model that achieves over 95% accuracy on clear recordings.

How Speechyou Helps

Speechyou offers a straightforward solution for Q'anjob'al transcription:

• Upload audio or video files (MP3, MP4, WAV, etc.)
• Select Q'anjob'al and optionally specify the dialect
• Receive a timestamped transcript in plain text, SRT, or VTT format
• Edit and export directly from the browser

Use cases include:

• Podcasters producing content in Q'anjob'al can generate show notes and subtitles.
• Teachers can transcribe Q'anjob'al language lessons for students.
• Researchers can quickly process field recordings.
• Community organizations can caption public service announcements.

Getting Started

To try Q'anjob'al speech to text, simply sign up for Speechyou's Solo plan, which includes unlimited transcription. No credit card required for the trial. Your recordings are processed securely, and you can download the results in minutes.

Preserve and promote Q'anjob'al with AI-powered transcription. Try Speechyou today.