Model Version

FluidAudio supports two model versions:

• v3
• v2

Choosing the correct model version impacts both language support and recognition accuracy.

v3 (Recommended)

• Supports 30 European languages
• Improved overall recognition accuracy
• Better noise robustness
• Enhanced diarization performance

v3 is recommended for:

• Multilingual users
• European language support
• Modern devices
• Most production environments

If you are not strictly English-only, v3 is the correct choice.

v2 (English Only)

• Tuned specifically for English
• Optimized English-language performance
• Slightly lower computational demand

v2 may be appropriate if:

• You operate exclusively in English.
• You are running on older hardware.
• You want a lightweight model with focused English tuning.

If you require non-English language support, v2 will not be suitable.