Ticuna Speech to Text: Preserving the Tones of the Amazon

Ticuna (Duüxü) is a language isolate spoken by the Ticuna people across the Amazon rainforest in Brazil, Colombia, and Peru. With an estimated 50,000 speakers, it is one of the largest indigenous languages in the Amazon, yet it remains under‐resourced in the digital world. Accurate speech‐to‐text technology for Ticuna is not just a convenience — it is a tool for language preservation, education, and cultural continuity.

Why Ticuna Transcription Matters

For decades, Ticuna was primarily oral. Missionaries developed a Latin‐script orthography in the 1970s, but literacy rates remain low, and most daily communication happens through speech. By converting spoken Ticuna into written text, Speechyou helps bridge the gap between oral tradition and digital literacy. Teachers can create subtitles for classroom videos, elders can record their stories for future generations, and linguists can analyse the language’s rich tonal system more easily.

The Challenge of Tones

Ticuna is a tonal language with five contrastive tones: high, mid, low, rising, and falling. A single syllable can change meaning depending on its pitch. For example:

• na (low tone) = water
• ná (high tone) = mother
• nà (falling tone) = to carry

This tonal complexity is a major challenge for off‐the‐shelf ASR systems. Most speech recognition tools are designed for non‐tonal languages and will confuse these words. Speechyou’s Ticuna model explicitly models pitch contours, using a combination of fundamental frequency tracking and neural network classifiers to distinguish each tone. The result is a transcription that preserves the meaning of the original speech.

Dialectal Variation

Ticuna speakers are spread across a wide geographic area, and three main dialects are recognised: Amazonas (Brazil‐Colombia border), Marajó (Brazilian island region), and Loreto (Peru). These dialects differ in vowel quality, tone realization, and vocabulary. Speechyou’s training data includes recordings from all three regions, and the model automatically adapts to dialectal features — no manual selection needed.

Use Cases for Ticuna Speech‐to‐Text

• Oral history preservation: Transcribe interviews with elders to create a written archive of traditional knowledge, mythologies, and medicinal plant uses.
• Bilingual education: Generate Ticuna‐Portuguese subtitles for school videos, helping children learn to read in both languages.
• Community media: Convert radio broadcasts and podcasts into text for distribution on social media and community websites.
• Linguistic research: Access time‐aligned transcriptions with tone labels for phonetic and phonological analysis.
• Accessibility: Provide live captions for Ticuna community meetings, making them inclusive for deaf or hard‐of‐hearing members who read Ticuna.
• Language revitalization: Create reading materials from spoken content, expanding the body of written Ticuna available to learners.

How Speechyou Handles Ticuna

Speechyou’s pipeline begins with a pre‐processor that segments the audio into short utterances. The acoustic model extracts features including Mel‐frequency cepstral coefficients and pitch contours. A transformer‐based decoder then generates the text, incorporating tone markers as diacritics on vowels. The system also supports code‐switching with Spanish and Portuguese, so bilingual recordings are transcribed accurately.

Users can upload audio or video files in common formats (MP3, WAV, MP4, etc.) or paste a YouTube link. The output can be downloaded as plain text, SRT, or VTT subtitles. For developers, a REST API allows batch processing of large corpora.

The Future of Ticuna ASR

As more Ticuna speakers use Speechyou, the model improves through active learning. Each new transcription adds to the training set, increasing accuracy for rare words, proper names, and dialectal variants. Speechyou is committed to working with Ticuna communities to ensure the technology respects their language and cultural context.

Transcribing Ticuna is more than a technical achievement — it is a step toward digital equality for one of the Amazon’s most important languages. With Speechyou, the voices of the Ticuna people can be heard in written form, preserved for generations to come.