CallMissed logoCallMissed
Back to all models
Speech to Textmultilingualstreamingbudget

Ink Whisper

by Cartesia · Released Undisclosed

Cartesia's fastest and most affordable speech-to-text model. 100 languages including Hindi, Urdu, and Tamil, with better accuracy than baseline Whisper and dynamic chunking that cuts hallucination during pauses. Works for both file transcription and live streaming at $0.18 per audio hour.

Speech to Text

Ink Whisper

Powered by Cartesia · Whisper-family ASR with dynamic chunking (Cartesia Ink)

Context Window

N/A

Parameters

Undisclosed

Max Output

N/A

Category

Speech to Text

Overview

Ink Whisper is Cartesia's fastest, most affordable speech-to-text model. It is a Whisper-family recognizer that Cartesia tuned for production transcription, delivering better accuracy than baseline Whisper while staying the cheapest STT option in Cartesia's lineup at $0.18 per audio hour.

Its most useful practical improvement is dynamic chunking. Baseline Whisper is known to hallucinate text during silence — long pauses and gaps in audio can produce confident but invented words. Ink Whisper adjusts how it segments incoming audio so those pauses and gaps produce far less spurious output, which matters most on real call recordings where hold music, dead air, and hesitation are normal.

The model covers 100 languages, including Hindi, Urdu, and Tamil, so it is a genuine multilingual option rather than an English-first model with partial coverage. Language can be passed explicitly when the channel is known, which reduces errors versus autodetect.

It is available on both CallMissed surfaces: batch file transcription over POST /v1/audio/transcriptions, and live streaming on a voice session. That makes it the Cartesia STT model to reach for when you need one id that works for both recorded audio and real-time transcription — its sibling ink-2 is streaming-only and English-only.

Choose Ink Whisper for cost-sensitive multilingual transcription: bulk call-recording archives, meeting and podcast transcription, Indic-language transcription pipelines, and voice agents where price per hour matters more than turn-detection sophistication.

Pricing

MetricPrice
Price /hour₹18.0000

1 credit = ₹1 = $0.01 USD. Transparent per-call pricing — you pay only for what you call, with no seat fees or minimums.

Key Highlights

  • 100 languages including Hindi, Urdu, and Tamil
  • Cartesia's fastest and most affordable STT
  • Better accuracy than baseline Whisper
  • Dynamic chunking cuts hallucination during pauses and gaps

Benchmarks

BenchmarkScore
Languages100
Price$0.18
AccuracyAbove baseline Whisper
DeploymentBatch + Streaming

Technical Details

  • Supports 100 languages, including Hindi, Urdu, and Tamil
  • Dynamic chunking reduces hallucination during pauses and audio gaps
  • Available for batch file transcription and live streaming
  • Cartesia's fastest and lowest-cost speech-to-text model
  • Optional language hint skips autodetect when the channel is known

Strengths

  • Cheapest Cartesia STT at $0.18 per audio hour
  • 100-language coverage including major Indic languages
  • Less hallucination on silence than baseline Whisper
  • Works for both file transcription and live streaming

Limitations

  • Paid plans only (not in the free tier)
  • No model-native turn detection — use ink-2 for turn-aware voice agents
  • Accuracy varies across the long tail of its 100 supported languages

Use Cases

Bulk call-recording transcriptionMeeting and podcast transcriptionIndic-language transcription pipelinesCost-sensitive voice agent STT

API Example

curl https://api.callmissed.com/v1/audio/transcriptions \
  -H "Authorization: Bearer cm_YOUR_KEY" \
  -F file=@call.mp3 \
  -F model=ink-whisper \
  -F language=hi

Endpoint: POST /v1/audio/transcriptions · Model ID: ink-whisper

Try Ink Whisper now

Get 1000 free API credits on signup. No credit card required.

Start free Book a DemoRead docs