Hume AI alternatives for emotion measurement

The alternatives, measured

Accuracy is seven-class emotion accuracy on speech-emotion-bench — open models on the full 64,384-clip evaluation; closed/API and audio-LLM systems on a fixed 5,000-clip stratified subset. All rows use the same seven-class mapping and scorer. Disclosure: this is a vendor page; the full 64-system results and protocol are published and downloadable so you can check our work.

Oruk Resonance 1 was trained in-distribution; the other measured systems were evaluated zero-shot across corpora. Their training exposure and evaluation sample sizes differ. The 77.8% result describes the July 2026 checkpoint, not Resonance 2, Fourier or current API accuracy.

How to choose

Hume Tagger / Prosody (current)

Hume currently advertises Tagger for offline batch analysis across 600+ expression dimensions and Prosody for real-time signals. Hume directs prospective Tagger users to contact Research. These current products were not evaluated in the July 2026 oruk benchmark; the 49.6% row belongs to the legacy 48-dimension prosody endpoint.

oruk Speech API

Resonance 1 returns timed English file analysis; Spectra-2 returns a final transcript and clip-level scores in 25 transcription languages. Resonance 2 provides acoustic scores, and Realtime provides live transcripts with variable-length phrase-emotion arrays. These are separate contracts. The displayed benchmark score belongs only to the July 2026 Resonance 1 checkpoint, trained in-distribution.

emotion2vec+ (self-hosted)

The strongest open-weights option measured. Free to run, but you own GPU serving, calibration, thresholds, and scaling. A good fit for teams with ML infrastructure.

Gemini 3 Flash Preview

The strongest frontier multimodal API measured. Prompt-based rather than calibrated scores; useful when emotion is a secondary need inside a broader LLM workflow.

Behavioral Signals

Speech analytics vendor with an emotion/behavior API aimed at call centers.

GPT-Audio 1.5

OpenAI’s audio-capable model, prompt-based. Same trade-off as Gemini: flexible, not calibrated.

audEERING devAIce

Long-standing paralinguistics vendor (makers of openSMILE); offers on-prem and embedded deployments.

Empath

Japanese vocal-emotion vendor; scored lowest of the commercial APIs measured.

The legacy Hume product also covered face, text, and vocal-burst modalities. Hume’s current public Expression Measurement page describes voice Tagger and Prosody products; it does not document those former modalities. If they are requirements, confirm current coverage directly with Hume or evaluate a multimodal specialist.

Try the integration on one recording

Start with the no-account analyzer, then use the English Resonance 1 Python recording walkthrough to inspect selected scores and timed segments. If you are replacing the legacy Hume response format, follow the migration guide; the labels and response contracts are not interchangeable.

Already have legacy batch predictions? The local response comparison tool matches overlapping intervals in your saved Hume and Oruk responses, preserving original scores, errors, and unmatched passages. It does not convert thresholds or test current Hume products.

The current API evaluation includes saved responses and an offline replay. It measures Fourier on a narrow CREMA-D subset, not current Hume Tagger or Prosody. Use the same consented recordings and a declared label mapping for your own comparison.

For an existing voice-agent stack, the Oruk and Valence Pulse comparison covers live transcription, phrase emotion and the different integration contracts. It also explains what a matched evaluation would need to measure.

FAQ

Does Hume still offer emotion measurement?

Yes. Hume’s current product page advertises Tagger for offline batch analysis and Prosody for real-time signals. The dated 49.6% legacy prosody snapshot in our benchmark is not evidence about either current product. Current Tagger access is contact-led according to Hume’s page.

Which option had the top measured result?

In oruk’s July 2026 speech-emotion-bench release, oruk Resonance 1 measured 77.8% seven-class accuracy. Oruk was trained in-distribution; the other measured systems were evaluated zero-shot across corpora. Open models were evaluated on the full 64,384 clips; closed/API and audio-LLM systems used a fixed 5,000-clip subset. This is not a measurement of Resonance 2, Fourier or current API accuracy. Current Hume Tagger and Prosody were not tested. The 49.6% Hume row is a legacy prosody snapshot.

Is there a free alternative to Hume Expression Measurement?

Open models such as emotion2vec+ can be self-hosted, subject to their license and your compute, serving and maintenance costs. Evaluate their outputs on your workload. Standard Oruk self-serve plans offer a 7-day trial with a card required; usage limits and normal subscription terms apply.

Sources and scope

• Hume AI’s current Expression Measurement page, reviewed October 3, 2026 — Tagger offline, Prosody real time, 600+ dimensions, 50+ languages, and contact-led Tagger access.
• Hume’s EVI and TTS notice, reviewed October 3, 2026 — sunset scope, exact access cutoff and account-data deletion.
• oruk benchmark methodology and limitations — sample split, in-distribution disclosure, downloadable results, and the dated legacy Hume snapshot.
• oruk is not affiliated with Hume AI. Report a factual issue to access@oruk.ai.