# The Cascade Equivalence Hypothesis: When Do Speech LLMs Behave Like ASRrightarrowLLM Pipelines?

arXiv: https://arxiv.org/abs/2602.17598 · HTML: https://mlsanity.com/paper/2602.17598/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 32 references that are in the map.
- First posted on arXiv 19 Feb 2026.

## Builds on (most cited first, 23 of 23)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://mlsanity.com/paper/2006.11477/) (2020)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [Character-level Convolutional Networks for Text Classification](https://mlsanity.com/paper/1509.01626/) (2015)
- [Understanding intermediate layers using linear classifier probes](https://mlsanity.com/paper/1610.01644/) (2016)
- [MELD: A Multimodal Multi-Party Dataset for Emotion Recognition in Conversations](https://mlsanity.com/paper/1810.02508/) (2018)
- [Eliciting Latent Predictions from Transformers with the Tuned Lens](https://mlsanity.com/paper/2303.08112/) (2023)
- [Qwen2-Audio Technical Report](https://mlsanity.com/paper/2407.10759/) (2024)
- [SpeechGPT: Empowering Large Language Models with Intrinsic Cross-Modal Conversational Abilities](https://mlsanity.com/paper/2305.11000/) (2023)
- [Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs](https://mlsanity.com/paper/2503.01743/) (2025)
- [MUSAN: A Music, Speech, and Noise Corpus](https://mlsanity.com/paper/1510.08484/) (2015)
- [LEACE: Perfect linear concept erasure in closed form](https://mlsanity.com/paper/2306.03819/) (2023)
- [AIR-Bench: Benchmarking Large Audio-Language Models via Generative Comprehension](https://mlsanity.com/paper/2402.07729/) (2024)
- [Towards Multimodal Sarcasm Detection (An _Obviously_ Perfect Paper)](https://mlsanity.com/paper/1906.01815/) (2019)
- [WavLLM: Towards Robust and Adaptive Speech Large Language Model](https://mlsanity.com/paper/2404.00656/) (2024)
- [Dynamic-SUPERB: Towards A Dynamic, Collaborative, and Comprehensive Instruction-Tuning Benchmark for Speech](https://mlsanity.com/paper/2309.09510/) (2023)
- [Scaling Properties of Speech Language Models](https://mlsanity.com/paper/2404.00685/) (2024)
- [Do Audio LLMs Really LISTEN, or Just Transcribe? Measuring Lexical vs. Acoustic Emotion Cues Reliance](https://mlsanity.com/paper/2510.10444/) (2025)
- [AudioLens: A Closer Look at Auditory Attribute Perception of Large Audio-Language Models](https://mlsanity.com/paper/2506.05140/) (2025)
- [Closing the Gap Between Text and Speech Understanding in LLMs](https://mlsanity.com/paper/2510.13632/) (2025)
- [Improving Causal Interventions in Amnesic Probing with Mean Projection or LEACE](https://mlsanity.com/paper/2506.11673/) (2025)

## Built on by (most cited first, 5 of 5)

- [Generating Synthetic Doctor-Patient Conversations for Long-form Audio Summarization](https://mlsanity.com/paper/2604.06138/) (2026)
- [WASIL: In-the-Wild Arabic Spoken Interactions with LLMs](https://mlsanity.com/paper/2605.16364/) (2026)
- [Interleaved Speech Language Models Latently Work In Text](https://mlsanity.com/paper/2606.22473/) (2026)
- [Inherited Heads: Audio language models track speakers with their text backbone's attention, and an attention-mass ranking retrieves a different set](https://mlsanity.com/paper/2609.14174/) (2026)
- [Transcribe, Then Reason: Two-Pass Decomposition for Multimodal Review](https://mlsanity.com/paper/2609.18958/) (2026)

## Abstract (arXiv)

Speech LLMs are widely understood to be better than ASRrightarrowLLM cascades since they have access to the audio directly, and not just the transcript. In this paper, we present an evaluation methodology and a mechanistic interpretation of the observed behavior of speech LLMs. First, we introduce matched-backbone testing which separates out the behavior of the speech LLM from the reasoning capabilities of the underlying LLM. Second, we provide a mechanistic analysis of speech LLMs using logit lens and LEACE and show the literal transcript emerging from the LLM's hidden states and that text representations are causally necessary. We also show that in most deployed use cases, current speech LLMs are expensive cascades, and under noise, they are worse ones, with clean-condition advantages reversing by up to 7.6% at 0dB.