Oruk
Summary
Most transcription APIs treat emotion as a binary sentiment score bolted on as an afterthought — you get 'positive' or 'negative' and you're on your own. oruk is a file-based Speech API built specifically for the gap between 'I have audio' and 'I need to know what was said, how it was said, and what emotional state it carried.'
The API processes prerecorded English audio files and returns transcripts, up to 15 multilabel emotion scores, 16 speaking-style labels, and time-local segments — all in a single POST call if you use the unified endpoint. The vendor's published benchmarks show the lowest word-error rate in their measured panel and a meaningful accuracy gap over the next-best open model on a 7-class emotion task. That benchmark lead is English-only, file-based, and self-reported — real-world audio with accents or background noise deserves your own held-out test set before you commit. Streaming is not supported; teams that need live transcription or real-time call analysis will hit a hard wall immediately.
Bottom line: oruk is a defensible choice for offline English speech pipelines that need calibrated emotion and style labels alongside a transcript — but if your use case requires streaming audio, non-English languages, or real-time inference, this API ends at the scope statement.
Pricing Plans
Usage-Based- Free Tier
- $50 trial credit
Resonance
Flagship model for transcription and unified analysis
- Transcription $0.0080 per audio min
- Emotion + style $0.0100 per audio min
- Unified analysis $0.0120 per audio min
Spectra 1
Efficient model for emotion and style
- Transcription $0.0045 per audio min
- Emotion + style $0.0075 per audio min
- Unified analysis $0.0090 per audio min
View full pricing on oruk.ai →
Pricing may have changed since last verified. Check the official site for current plans.
Community Performance Report Card
No community ratings yet. Be the first to rate this tool!
Community Benchmarks Community
Sign in to submit a benchmarkNo community benchmarks yet. Be the first to share a real-world data point.
Pros
Sign in to edit- Multilabel emotion output with calibrated scores across 15 classes, so downstream systems can act on co-occurring emotional states rather than forcing a single label onto ambiguous audio.
- Unified analysis endpoint returns transcript, emotion labels, style labels, and time-local segments in one request, which means teams avoid building and maintaining a chained multi-call pipeline to get the same data.
- Provider benchmarks show the lowest measured word-error rate in their evaluated panel, so teams replacing Whisper or Azure Speech for English transcription accuracy have a published comparison point to test against.
- Affect endpoint skips transcript generation when only emotion and style scores are needed, which reduces per-request cost and latency for pipelines where the transcript already exists.
- API access requires no card to start, so teams can run evaluation against their own audio before committing to production billing.
Cons
Sign in to edit- The API is English-only with no multilingual support in the current scope statement. Teams processing Spanish, French, German, or any other language have no path forward here and will need to evaluate alternatives such as Deepgram or AssemblyAI from the start.
- Streaming is not supported — the contract is file-based only. Any team building a real-time call analysis product, a live transcription overlay, or a latency-sensitive voice interface hits this ceiling on day one and has to switch to a different provider entirely.
- Spectra 2, the next model tier listed in the catalog, is not yet serving traffic. Teams who plan a roadmap dependency on that model are blocked until the vendor announces general availability, with no timeline published on the vendor page.
- No self-hosted option exists, so teams with data residency requirements, air-gapped environments, or strict audio data retention policies cannot use this API without routing audio through the vendor's infrastructure.
Community Reviews
Sign in to write a reviewNo reviews yet. Be the first to share your experience.
About
- Platforms
- Web API
- API Available
- Yes
- Self-Hosted
- No
- Last Updated
- 2026-07-26T08:22:34.926Z
Best For
Who it's for
- Applications needing emotion and style labels alongside transcripts
- English-only speech processing workflows
- Developers requiring calibrated multilabel outputs
What it does well
- English audio transcription
- Multilabel emotion detection on speech
- Speaking-style classification
- Unified audio analysis with segments and tags
Discussion Community
Sign in to commentNo discussion yet. Sign in to start the conversation.
Compare Oruk
Spotted incorrect or missing data? Join our community of contributors.
Sign Up to ContributeCommunity Notes & Tips Community
Sign in to contributeBe the first to contribute. General notes, observations, gotchas, and tips from people who use this tool day-to-day.
Frequently Asked Questions
- Is Oruk free?
- Oruk has a permanent free tier alongside paid upgrades. You can keep using a baseline version indefinitely without paying.
- Is Oruk open source?
- No — Oruk is a closed-source tool. Source code is not publicly available.
- Does Oruk have an API?
- Yes. Oruk exposes a developer API. See the official documentation at https://oruk.ai for details.
- When was Oruk released?
- Oruk was first released in 2026.
- What platforms does Oruk support?
- Oruk is available on: Web API.
Hours Saved & ROI Stories Community
Sign in to contributeBe the first to contribute. Concrete time/cost savings, with context. e.g. "Cut my code review backlog from 4h to 45m per week."
Curated lists that include this category
oruk is a file-based Speech API that accepts prerecorded English audio and returns transcription, emotion detection, speaking-style classification, or a unified response combining all three. The core workflow is a POST request against one of three endpoints — transcription, affect, or unified analysis — with the model tier specified in the request. The unified endpoint returns a transcript, calibrated multilabel scores for emotion and style, time-local segments, and tagged text in a single response, which means teams building downstream annotation or review tooling get everything in one round trip rather than chaining separate calls.
The differentiating feature is multilabel calibration on held-out audio. Rather than outputting a single dominant emotion class, the API returns scores for up to 15 emotion labels simultaneously — a clip can carry fear and surprise at the same time, and the scores are calibrated against out-of-sample data rather than raw softmax outputs. The vendor’s July 2026 benchmark report shows per-class F1 scores across all seven mapped emotion classes, with the Spectra model posting results materially above Gemini 3 Flash Preview on disgust, fear, surprise, and anger specifically. These figures come from the vendor’s own evaluation methodology; the report documents its limitations and the dataset sources used.
Two models are actively serving traffic: Resonance, described as the flagship for highest-accuracy transcription and unified analysis, and Spectra 1, positioned for efficient affect workloads where transcript output can be skipped. A third model, Spectra 2, is listed in the catalog as a preview tier but the docs state it is not yet serving inference traffic. The scope is deliberately narrow — English, file-based, prerecorded audio only. No self-hosted deployment is available, and the vendor page shows no streaming contract in the current API v1 specification.