We use Nova 2 or Nova 3 transcribers for Speech-to-Text when transcribing conversations in Turkish. However, we are not satisfied with the performance of these transcribers. We often have difficulty understanding what is said. Do you have any recommendations for a transcript provider, engine, or model that will work flawlessly in Turkish?
Thank you so much!
s
Shaunak
04/08/2026, 6:52 AM
Hi, thanks for reaching out.
Shaunak
04/08/2026, 6:52 AM
For Turkish transcription, you could try Speechmatics with the enhanced model. It's designed for high accuracy across 55+ languages, including Turkish, and supports custom vocabulary for domain-specific terms.
Shaunak
04/08/2026, 6:52 AM
If Speechmatics doesn't work well, other options to try:
1. Google STT with multilingual mode
2. Gladia with automatic language detection
3. Azure Speech Services (if you have a custom API key)
You can test these directly in the dashboard using the Test Assistant feature to compare performance before committing to one.
m
Matt
04/08/2026, 7:51 AM
Totally get this, Turkish can be tough, and Nova isn’t always great with it. You’ll likely get better accuracy with Whisper (large-v3) or Google Speech-to-Text, especially with some audio cleanup. Happy to help you test and set this up properly.
are your recordings clean, or do they have multiple speakers/noise? @muratsk
m
muratsk
04/09/2026, 10:57 AM
@Shaunak thanks my friend, it helps a lot for me! Speechmatics really sounds cool when i tried for few calls.
s
Shaunak
04/14/2026, 7:55 AM
no worries!
Shaunak
04/14/2026, 7:56 AM
if you face any further issues, feel free to open a new ticket.