Mazandarani Speech to Text: Preserving a Language Through AI

The Mazandarani Language and Its Speakers

Mazandarani (مازرونی), also known as Tabari, is a Northwestern Iranian language spoken primarily in the Mazandaran province of Iran, along the southern coast of the Caspian Sea. With an estimated 3 to 4 million native speakers, it is one of the major regional languages of Iran, yet it remains underrepresented in digital tools. The language has its own rich literary tradition, including epic poetry and folk tales, but in daily life it often competes with Persian, the national language. Many Mazandarani speakers are bilingual, switching between Mazandarani and Persian depending on context.

Why Accurate Speech-to-Text Matters for Mazandarani

Accurate speech-to-text for Mazandarani is not just a convenience; it is a tool for language preservation and accessibility. Here’s why it matters:

• Documenting oral traditions: Elder speakers hold a wealth of stories, songs, and proverbs that may otherwise be lost. Transcription makes them searchable and shareable.
• Education: Schools in Mazandaran are increasingly using Mazandarani in early childhood education. Teachers need transcripts for lesson planning and assessment.
• Media and content creation: Local radio, podcasts, and YouTube channels in Mazandarani can benefit from subtitles to reach a wider audience, including the deaf community.
• Research: Linguists studying Iranian languages rely on accurate transcriptions for phonetic and grammatical analysis.

Challenges in Transcribing Mazandarani

Transcribing Mazandarani with AI presents several challenges:

• Vowel reduction and harmony: Unstressed vowels are often shortened or merged, making it hard for generic ASR to distinguish words like "بورده" (borde, to go) from "بورده" (borde, to carry) without context.
• Dialect diversity: The four main dialect groups (Sari, Babol, Amol, Gorgani) differ in pronunciation, lexicon, and even some grammatical forms. A model trained on Sari may struggle with Gorgani.
• Code-switching: Speakers frequently insert Persian words, especially for modern concepts. A transcription system must recognize when the language changes and apply appropriate rules.
• Lack of labeled data: Unlike major languages, Mazandarani has few publicly available transcribed speech datasets. Speechyou overcomes this by using transfer learning and community-contributed recordings.

How Speechyou Handles These Challenges

Speechyou’s Mazandarani model is built from the ground up to address these issues:

• Dialect-aware training: We include audio samples from all major dialects, and users can select a dialect profile for improved accuracy.
• Multilingual engine: The same model processes Persian and Mazandarani, with language identification at the word level to handle code-switching.
• Custom acoustic features: The model learns the unique vowel reduction patterns and consonant clusters of Mazandarani, such as the retroflex sounds found in some dialects.
• Continuous improvement: Users can submit corrections, which are used to retrain the model, making it smarter over time.

Use Cases for Mazandarani Transcription

Podcasts and Audio Content

Local podcasters like "Mazandaran Podcast" can automatically transcribe episodes, turning spoken content into searchable text for show notes or blog posts. This increases discoverability and helps non-fluent speakers follow along.

Subtitle Generation for Videos

A YouTuber creating content about Mazandarani cuisine can generate SRT subtitles in Mazandarani, then translate them into English or Persian to reach a global audience. Speechyou’s subtitle editor allows easy timing adjustments.

Academic Research

Linguists at the University of Mazandaran can transcribe field recordings of elder speakers, preserving dialectal variations for future study. The ability to export in plain text or SRT makes integration with annotation tools seamless.

Accessibility

Community events, religious gatherings, and local TV programs can provide real-time captions in Mazandarani for deaf viewers. Speechyou’s low-latency streaming transcription makes this possible.

Oral History Preservation

Families can record interviews with grandparents and transcribe them, creating a written record of family history in the native language. These transcripts can be stored, printed, or shared with relatives abroad.

Getting Started with Speechyou for Mazandarani

Using Speechyou is simple:

1. Upload your audio or video file (or paste a YouTube link).
2. Select "Mazandarani (Latin script)" or "Mazandarani (Perso-Arabic script)" as the output language.
3. Optionally, choose a dialect profile if you know the speaker’s region.
4. Click transcribe and receive your text in minutes.
5. Export as SRT, VTT, plain text, or use the built-in translation feature to add subtitles in other languages.

Speechyou is available on web and mobile, with a free tier for testing. The Solo plan includes unlimited transcription, making it affordable for individual creators and researchers.

The Future of Mazandarani in the Digital Age

As Mazandarani speakers continue to produce digital content, the need for AI tools that understand their language will only grow. Speechyou is committed to supporting minority languages like Mazandarani, ensuring that no language is left behind in the AI revolution. By providing accurate, dialect-aware transcription, we help preserve the linguistic heritage of the Caspian region while making it accessible to the world.