Whisper large-v3-turbo
Fast speech-to-text in 99 languages, Sinhala and Tamil among them.
About Whisper large-v3-turbo
Whisper large-v3-turbo turns speech into text in 99 languages. It is a slimmed-down Whisper large-v3: OpenAI cut its decoder from 32 layers to 4 and retrained it, making it much faster for accuracy roughly on par with the earlier large-v2. It is the default model in OpenAI’s open-source Whisper today.
Use it for meeting and call transcripts, subtitles and voice notes. Like every Whisper, it can occasionally write words that were never said, especially over silence or noise, so keep a person checking transcripts that matter.
Where it’s strong
Meeting and call transcripts.
Subtitles and captions.
Voice notes and voice input in apps.
Where to be careful
Not built for translating speech into English — OpenAI left translation out of its training.
Can invent text that is not in the audio, and accuracy drops on strong accents and less common languages.
Price in rupees
Every rate is published and stays the same from one request to the next. See every model.
Specs
whisper-large-v3-turboUsing it from Sri Lanka
Whisper large-v3-turbo is billed in rupees, on the same monthly invoice as every other model your team uses — no foreign card, no separate OpenAI account, and one key for all of it.
Prompts sent to Whisper large-v3-turbo are processed outside Sri Lanka. If a workload has to stay in the country, Qwen3.8-27B (Colombo) runs on our own servers in Colombo.
Sinhala and Tamil are both on Whisper’s official language list, so it will transcribe them. OpenAI publishes no accuracy figures for either and says quality is lower on languages with less training data — test it on your own recordings.
Call it
Use the OpenAI SDK you already have — point it at Roar AI and name the model. Read the docs.
from openai import OpenAI
client = OpenAI(base_url="https://api.roar-ai.com/v1", api_key="roar_live_…")
text = client.audio.transcriptions.create(
model="whisper-large-v3-turbo",
file=open("call-recording.mp3", "rb"),
)Questions
How much does Whisper large-v3-turbo cost in Sri Lanka?
Price is $14.4444 per 1M audio seconds. The rate is published and does not change from one request to the next.
Can I pay for Whisper large-v3-turbo in Sri Lankan rupees?
Yes. Usage is billed to one monthly invoice in rupees, together with every other model your team uses, and you can also pay by card. There is no separate OpenAI account to open and no foreign card needed.
Does Whisper large-v3-turbo keep data in Sri Lanka?
No — prompts sent to Whisper large-v3-turbo are processed outside Sri Lanka. If your data has to stay in the country, use Qwen3.8-27B (Colombo), which runs on our servers in Colombo.
Is Whisper large-v3-turbo open source?
Its weights are open under the MIT licence, so you could run it yourself. Through Roar AI you call it like any other model, with no servers to manage.