Nova-3 Adds 12 New Languages Across Europe and South Asia

Deepgram has added support for 12 new monolingual languages to Nova-3, now available in production. This release brings Nova-3 into additional regions with diverse linguistic characteristics, delivering Nova-3–level accuracy, Keyterm Prompting, and production-ready performance to developers and enterprises building global voice applications.

Deepgram continues to broaden Nova-3’s language coverage, adding support for languages spoken across Southeastern Europe and South Asia. These regions include languages with distinct phonetic systems, grammatical structures, and scripts that have historically posed challenges for general purpose speech recognition systems

This release further establishes Nova-3 as an enterprise-grade speech-to-text model, delivering high accuracy and reliable performance across a wider range of languages, scripts, and speech patterns.

The following languages are now supported on Nova-3:

Southeastern Europe

• Bosnian - (bs)
• Croatian - (hr)
• Macedonian - (mk)
• Serbian - (sr)
• Slovenian - (sl)
• Belarusian - (be)

Languages in this region often feature rich inflection, flexible word order, and complex morphology. Nova-3 is designed to handle these characteristics while maintaining low latency and consistent transcription quality across both batch and streaming workflows.

South Asia

• Bengali - (bn)
• Marathi - (mr)
• Tamil - (ta)
• Telugu - (te)
• Kannada - (kn)
• Tagalog - (tl)

South Asian languages span multiple language families and writing systems, with significant variation in phonetics and sentence structure. Nova-3 supports these differences natively, enabling accurate transcription for conversational, enterprise, and analytics-driven use cases.

Earlier Nova-3 expansions focused on widely spoken European and Asian languages. This release represents the next phase, extending Nova-3 into regions with greater linguistic diversity and structural complexity.

Nova-3 adapts to the unique characteristics of each language by supporting:

• Languages with complex morphology and inflection
• Multiple scripts and writing systems
• Regional pronunciation and speech variation

All while preserving the performance, reliability, and scalability required for production deployments.

These newly supported languages also benefit from Keyterm Prompting on Nova-3.

Keyterm Prompting allows developers to guide transcription toward domain-specific terminology, jargon, brand names, and keywords, improving recognition without retraining models or managing custom vocabularies.

This capability is especially valuable for:

• Call centers and customer support systems
• Voice agents and IVR applications
• Industry-specific analytics and transcription workflows

Key terms are applied dynamically at infereence time, making customization fast and flexible across languages.

As products expand into new markets, speech recognition systems must adapt to local languages, scripts, and speech patterns, without increasing system complexity.

With this release, Nova-3 continues to evolve as a unified speech-to-text foundation for global applications. Rather than applying a single pattern to every language, Nova-3 supports diverse linguistic structures, whether that involves complex morphology, multiple scripts, or regional speech variation.

For developers and enterprise teams, this means:

• Consistent performance across diverse global regions
• Reliable transcription in both conversational and formal speech
• Production-ready performance for real-time and batch workflows
• Flexible customization with Keyterm Prompting for domain-specific accuracy
• Lower latency and fewer transcription errors in multilingual environments
• A single API for scaling speech-to-text globally

All 12 languages are available through the same API developers already use today. You can test all supported languages directly in the Deepgram Playground before deploying to production.

Switching to any of the newly supported languages is simple. Update your API request with the appropriate language code:

curl --request POST \

--header "Authorization: Token YOUR_DEEPGRAM_API_KEY" \

--header "Content-Type: audio/wav" \

--data-binary @youraudio.wav \

"https://api.deepgram.com/v1/listen?model=nova-3&language=hr"

Supported language codes: 

hr, bs, mk, sr, sl, be, bn, mr, ta, te, kn, tl

Sign up free and unlock $200 in credits, enough to power over 750 hours of transcription or 200 hours of speech-to-text across Nova-3’s growing language suite. Explore details on our Models & Languages Overview page and experience Nova-3’s world-class adaptability for yourself.