Skip to content
Get started free

Explore oruk

Call analytics

AI call analytics

Turn recorded English calls into structured data: transcript and selected emotion and speaking-style scores in timed acoustic segments. One synchronous REST call per recording, using your plan’s audio-minute allowance.

Plans from $9/month · 7-day free trial · minutes included

Create your free account

Choose a plan after signup. Self-serve plans include a 7-day free trial.

Test your voice first — no account

Already have an account? Sign in · Compare plans

curl --fail-with-body https://speech-api.oruk.ai/v1/audio/analysis \
  -H "Authorization: Bearer $ORUK_API_KEY" \
  -F file=@call.wav -F model=oruk-resonance

Illustrative response · values depend on the recording

{
  "model": "oruk-resonance",
  "text": "I need help with this order.",
  "emotions": [{ "label": "frustrated", "score": 0.83 }],
  "styles": [{ "label": "confident", "score": 0.72 }],
  "usage": { "audio_seconds": 4.2, "billable_seconds": 4.2 }
}

Choose a packaged application or build with an API

Start with who will use the result. A review team may need a ready-made workspace with recording imports, access controls and reporting. A product team may need transcripts and timed annotations inside an existing application. Oruk supplies the analysis for that second path; your team builds the surrounding workflow.

Choose an application when

Reviewers need to start using a complete tool. Evaluate its recording integrations, permissions, search, exports and review process against the way your team works.

Choose an API when

You already control the recordings and application experience. Plan for server-side uploads, response storage, playback, reviewer access and the actions your product takes after a review.

From a completed recording to a review

This is an application design for the synchronous file API. Your application queues recordings and tracks job state; Oruk returns the analysis in the HTTP response. It does not create the review queue or a dashboard for you.

  1. 1. Receive the recording

    Keep your call ID and access rules with the recording in your own storage. Check that the file is English audio within the 30 MB and 60-minute limits before scheduling an upload. Keep API keys on your server.

  2. 2. Request unified analysis

    Send the file to POST /v1/audio/analysis with model=oruk-resonance. Add diarize=true when you need speaker turns. Speaker labels belong to that recording; they do not identify a customer or an agent. Keep your request ID for tracing failures.

  3. 3. Store the returned result

    Associate the JSON response with your call ID and recording. Preserve segment timestamps, returned labels, usage and any model-version fields. Oruk’s inference services discard audio and outputs after responding. Optional speaker diarization retains uploaded audio for up to 48 hours and speaker-label results for up to 24 hours. Your application controls storage and deletion of its copies. Handle failed requests separately from completed reviews.

  4. 4. Let a reviewer hear the passage

    Show the transcript beside playback at the returned timestamps. Treat emotion and style scores as annotations to inspect, not probabilities of a person’s feelings. Let reviewers record whether a passage matters and why before using the result in a QA report or CRM update.

The Python call-review guide provides a runnable starting point. The worked conversation example shows how to read four passages from a saved response, including multiple labels on one reply and the limits of comparing different speakers.

Use the production guide to set timeouts, bounded retries and usage monitoring. Check data handling and access requirements before choosing what your application stores.

Where teams apply it

You can use the endpoint in these call-review workflows: contact center analytics for queue-wide scoring and QA, voice of customer analytics for trending how customers sound over time, sales call analysis for coaching against specific moments in a deal, and voice call sentiment analysis for the timed emotion and speaking-style results themselves.

Evaluate the review workflow on your calls

Start with English calls that reflect your accents, microphones, background noise and call lengths. Have reviewers identify useful passages before showing the model annotations, then record what the annotations add or miss. Measure review time, missed passages and unnecessary flags alongside request failures and audio usage. These are evaluation measures to collect, not results Oruk has established for your team.

The published speech-emotion benchmarks are historical model comparisons, with an in-distribution Oruk entry. Their seven-class results do not establish the current endpoint’s accuracy on your calls. Read the current-API evaluation requirements before treating a result as evidence for deployment.

Plan the integration and its cost

Full emotion and speaking-style analysis uses completed English recordings. The separate Realtime preview provides live transcript tokens and phrase-level emotion. Start with the API docs for the request and response contracts, and the speech analytics API overview for integration scope.

With Resonance, one minute of audio uses one plan minute per request, including diarization. A unified request provides the transcript and acoustic labels together; separate calls process and meter the audio again. Use the pricing calculator and plan overage rates to estimate the workload. Reference cost fields in a response are not additional subscription charges.

For the score definitions, see the speech emotion API. You can explore a recording first in the voice emotion analyzer or try the captioning demo.

Start free trial

Plans from $9/month. $0 today, cancel any time.

Frequently asked questions

What is AI call analytics?
AI call analytics uses machine learning to turn recorded calls into structured data: what was said, who sounded how, and where the conversation turned. The oruk Speech API returns a transcript plus selected emotion and speaking-style scores for timed acoustic segments of a recording. Segment boundaries need not match sentences; use the timestamps to review how the call sounded alongside its words.
How is this different from call center analytics software?
Choose a packaged application when you need a ready-made review workspace, recording integrations and reporting. Choose an API when your team is building those workflows into its own product. Oruk accepts audio and returns JSON; your application manages recording access, storage, reviewer permissions, dashboards and CRM updates.
Does it work in real time during a call?
oruk-realtime streams multilingual transcript tokens and phrase-level emotion scores during a call. The richer 15-emotion, 16-speaking-style, tagged-transcript analysis remains a synchronous file request for post-call scoring and QA.
What audio does it accept?
Resonance accepts prerecorded English audio: WAV, FLAC, MP3, M4A, OGG, or WebM, up to 30 MB and 60 minutes per file. Standard self-serve plans start with a 7-day free trial: a card is required, $0 is charged today, and you can cancel before the trial ends. Promotional offers show their own terms at signup.
What signals come back for each call?
POST /v1/audio/analysis returns a transcript, selected scores from its 15-emotion and 16-style vocabularies, timed segments, a tagged transcript, and measured audio usage. Several labels may be returned, and styles can be empty. Subscription charges follow your plan allowance and overage rate, not reference cost fields.

Try it free for 7 days

Plans start at $9 a month, with access to every speech model. Add a card, pay nothing today, and cancel any time before the trial ends.

Start free trialRead the docs