Real-Time Voice AI Hears but Does Not Listen

• URL: http://arxiv.org/abs/2606.26083v1
• Date: Wed, 24 Jun 2026 17:55:38 GMT
• Title: Real-Time Voice AI Hears but Does Not Listen
• Abstract summary: We evaluate four leading production realtime voice systems.<n>All four systems act on the words rather than the voice.<n>We term this disconnect between perception and action the emotional intelligence gap of voice AI.
• Score: 28.50011328763975
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Speech conveys information through both words and vocal delivery. We evaluate four leading production realtime voice systems-OpenAI's GPT Realtime 2, Google's Gemini 3.1 Flash Live, and Alibaba's Qwen3.5 Omni Plus and Omni Flash-on tasks where the words and the delivery patterns both convey meaningful information. Across three consequential scenarios, all four systems act on the words rather than the voice. They end calls with crying callers who insist nothing is wrong, approve wire transfers authorized in frightened voices, and enroll callers whose agreement is clearly sarcastic. Surprisingly, this is often not a failure of perception. When asked directly, three of the four systems reliably identify the distress, fear, or sarcasm they later ignore when making decisions. We observe a similar pattern when these realtime voice systems estimate accent and age, as their responses frequently follow the biases of the words rather than the acoustic properties of the speaker. We term this disconnect between perception and action the emotional intelligence gap of voice AI. Prompting systems to explicitly attend to vocal delivery improves performance only partially and inconsistently. Our findings show that current realtime voice AI systems often behave as if speech had been reduced to a transcript, suggesting that they should be used with caution in settings where the tone and emotion of delivery convey important information.

Related papers

• Noise Adaptive Streaming Audio-Visual Speech Token Enhancement for Robust Full-Duplex Spoken Dialogue Models [49.372283081631366]
  AVSTE restores semantic speech tokens from noisy audio and lip video before they reach speech perception.<n>When integrated with Moshi, AVSTE improves average GPT-4ojudged response coherence from 1.42 1.91 under same-data-set speaker interference.
  arXiv  Detail & Related papers  (2026-09-08T07:59:13Z)
• Hear2Act: Benchmarking When Prosody Should Change What an Assistant Does [27.909684871924004]
  We introduce Hear2Act, a unified evaluation protocol for text and spoken assistants.<n>We evaluate decisions under transcript, audio, and concern-state access.<n>We show that prosody matters when lexical evidence is insufficient.
  arXiv  Detail & Related papers  (2026-08-20T00:16:36Z)
• From Speech to Interaction: Analyzing Multimodal Systems in Cocktail-Party Scenarios [68.14493981116665]
  Humans have the remarkable ability to selectively attend to individual speakers while filtering out competing speech from nearby conversations.<n>This "cocktail party" scenario still presents severe challenges to speech recognition systems.<n>The CHiME-9 MCoRec task provides a testbed where systems must recognize groups of speakers and transcribe each of their conversations from audio-visual input.
  arXiv  Detail & Related papers  (2026-08-09T06:11:38Z)
• RW-Voice-EQ Bench: A Real World Benchmark for Evaluating Voice AI Systems [5.434878413067881]
  We introduce the Real World Voice EQ Bench, a multidimensional benchmark for evaluating voice AI across text-to-speech (TTS), speech-to-speech (STS), speech understanding (SU), and automatic speech recognition (ASR)<n>Our evaluations indicate that performance is highly dimension-specific.
  arXiv  Detail & Related papers  (2026-07-16T11:15:16Z)
• Do AI Voices Learn Social Nuances? A Case of Politeness and Speech Rate [0.0]
  This study investigates whether state-of-the-art text-to-speech systems have the human tendency to reduce speech rate to convey politeness.<n>We prompted 22 synthetic voices from two leading AI platforms to read a fixed script under both "polite and formal" and "casual and informal" conditions.<n>Across both AI platforms, the polite prompt produced slower speech than the casual prompt with very large effect sizes.
  arXiv  Detail & Related papers  (2025-11-12T07:44:42Z)
• Chronological Thinking in Full-Duplex Spoken Dialogue Language Models [66.84843878538207]
  Chronological Thinking aims to improve response quality in full SDLMs.<n>No additional latency: once the user stops speaking, the agent halts thinking and begins speaking without further delay.<n>Results: Experiments demonstrate the effectiveness of chronological thinking through both objective metrics and human evaluations.
  arXiv  Detail & Related papers  (2025-10-02T10:28:11Z)
• Character-aware audio-visual subtitling in context [58.95580154761008]
  This paper presents an improved framework for character-aware audio-visual subtitling in TV shows. Our approach integrates speech recognition, speaker diarisation, and character recognition, utilising both audio and visual cues. We validate the method on a dataset with 12 TV shows, demonstrating superior performance in speaker diarisation and character recognition accuracy compared to existing approaches.
  arXiv  Detail & Related papers  (2024-10-14T20:27:34Z)
• Moshi: a speech-text foundation model for real-time dialogue [78.88479749811376]
  Current systems for spoken dialogue rely on pipelines independent voice activity detection and text-to-speech. We show how Moshi Moshi can provide streaming speech recognition and text-to-speech. Our resulting model is first real-time full spoken large language model modality.
  arXiv  Detail & Related papers  (2024-09-17T17:55:39Z)
• Towards Unsupervised Speech Recognition Without Pronunciation Models [57.222729245842054]
  In this article, we tackle the challenge of developing ASR systems without paired speech and text corpora.<n>We experimentally demonstrate that an unsupervised speech recognizer can emerge from joint speech-to-speech and text-to-text masked token-infilling.<n>This innovative model surpasses the performance of previous unsupervised ASR models under the lexicon-free setting.
  arXiv  Detail & Related papers  (2024-06-12T16:30:58Z)
• The timing bottleneck: Why timing and overlap are mission-critical for conversational user interfaces, speech recognition and dialogue systems [0.11470070927586018]
  We evaluate 5 major commercial ASR systems for their conversational and multilingual support. We find that word error rates for natural conversational data in 6 languages remain abysmal, and that overlap remains a key challenge. Our findings help to evaluate the current state of conversational ASR, contribute towards multidimensional error analysis and evaluation, and identify phenomena that need most attention on the way to build robust interactive speech technologies.
  arXiv  Detail & Related papers  (2023-07-28T11:38:05Z)
• A unified one-shot prosody and speaker conversion system with self-supervised discrete speech units [94.64927912924087]
  Existing systems ignore the correlation between prosody and language content, leading to degradation of naturalness in converted speech. We devise a cascaded modular system leveraging self-supervised discrete speech units as language representation. Experiments show that our system outperforms previous approaches in naturalness, intelligibility, speaker transferability, and prosody transferability.
  arXiv  Detail & Related papers  (2022-11-12T00:54:09Z)
• DualVoice: Speech Interaction that Discriminates between Normal and Whispered Voice Input [16.82591185507251]
  There is no easy way to distinguish between commands being issued and text required to be input in speech. The input of symbols and commands is also challenging because these may be misrecognized as text letters. This study proposes a speech interaction method called DualVoice, by which commands can be input in a whispered voice and letters in a normal voice.
  arXiv  Detail & Related papers  (2022-08-22T13:01:28Z)
• Self-Supervised Speech Representations Preserve Speech Characteristics while Anonymizing Voices [15.136348385992047]
  We train several voice conversion models using self-supervised speech representations. Converted voices retain a low word error rate within 1% of the original voice. Experiments on dysarthric speech data show that speech features relevant to articulation, prosody, phonation and phonology can be extracted from anonymized voices.
  arXiv  Detail & Related papers  (2022-04-04T17:48:01Z)
• Analysis and Tuning of a Voice Assistant System for Dysfluent Speech [7.233685721929227]
  Speech recognition systems do not generalize well to speech with dysfluencies such as sound or word repetitions, sound prolongations, or audible blocks. We show that by tuning the decoding parameters in an existing hybrid speech recognition system one can improve isWER by 24% (relative) for individuals with fluency disorders.
  arXiv  Detail & Related papers  (2021-06-18T20:58:34Z)
• Learning Explicit Prosody Models and Deep Speaker Embeddings for Atypical Voice Conversion [60.808838088376675]
  We propose a VC system with explicit prosodic modelling and deep speaker embedding learning. A prosody corrector takes in phoneme embeddings to infer typical phoneme duration and pitch values. A conversion model takes phoneme embeddings and typical prosody features as inputs to generate the converted speech.
  arXiv  Detail & Related papers  (2020-11-03T13:08:53Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.