Whisper-1

Technical Specifications of whisper-1

What is whisper-1?

whisper-1 is a speech recognition model available through CometAPI for turning audio into text and creating translations from spoken audio into English. It is designed for developers who need reliable transcription for recorded speech, interviews, meetings, voice notes, subtitles, and multilingual audio workflows.

The model is well suited for applications that need automatic speech recognition across multiple languages. It can transcribe audio in the original language or translate spoken content into English, making it useful for global products, media processing pipelines, support tools, and accessibility solutions.

Because whisper-1 works on uploaded audio files and returns text output, it fits naturally into backend automation, content indexing, caption generation, search enrichment, and analytics pipelines.

Main features of whisper-1

• Speech-to-text transcription: Converts spoken audio into written text for documents, captions, archives, and application workflows.
• Speech translation: Creates English text translations from non-English spoken audio, simplifying multilingual content processing.
• Multilingual recognition: Supports recognition across many languages, making it practical for international and cross-region deployments.
• Prompt-assisted formatting: Accepts prompts that can help guide punctuation, capitalization, terminology, and transcript style.
• File-based API workflow: Works well with uploaded audio files, making it easy to integrate into batch jobs, media systems, and backend services.
• Language identification support: Can be used in workflows where detecting or handling multiple spoken languages is important.
• Strong fit for content operations: Useful for subtitle generation, searchable transcript creation, customer call logging, interview processing, and voice-note conversion.

How to access and integrate whisper-1

Step 1: Sign Up for API Key

To start using whisper-1, first create an account on CometAPI and generate your API key from the dashboard. After logging in, go to the API management section, create a new key, and store it securely. This key will be required to authenticate every request you send to the whisper-1 API.

Step 2: Send Requests to whisper-1 API

Once you have your API key, you can send requests to the CometAPI endpoint using the whisper-1 model ID. Include your API key in the Authorization header and specify whisper-1 as the target model. For speech workflows, send an audio file to the appropriate transcription or translation endpoint.

curl --request POST \
  --url https://api.cometapi.com/v1/audio/transcriptions \
  --header "Authorization: Bearer YOUR_COMETAPI_KEY" \
  --header "Content-Type: multipart/form-data" \
  --form "model=whisper-1" \
  --form "file=@/path/to/audio.mp3"

For translation workflows, use the translation endpoint with the same model ID:

curl --request POST \
  --url https://api.cometapi.com/v1/audio/translations \
  --header "Authorization: Bearer YOUR_COMETAPI_KEY" \
  --header "Content-Type: multipart/form-data" \
  --form "model=whisper-1" \
  --form "file=@/path/to/audio.mp3"

Step 3: Retrieve and Verify Results

After the request is processed, CometAPI will return the generated text result for your whisper-1 job. Review the response to confirm transcript quality, language handling, punctuation, and completeness. If needed, refine your audio preprocessing or prompting approach and resend the request to improve output consistency for your production use case.