An official website of the United States government
Conversational systems rely heavily on speech recognition to interpret and respond to user commands and queries. Despite progress on speech recognition accuracy, errors may still sometimes occur and can significantly affect the end-user utility of such systems. While visual feedback can help detect errors, it may not always be practical, especially for people who are blind or low-vision. In this study, we investigate ways to improve error detection by manipulating the audio output of the transcribed text based on the recognizer's confidence level in its result. Our findings show that selectively slowing down the audio when the recognizer exhibited uncertainty led to a 12% relative increase in participants' ability to detect errors compared to uniformly slowing the audio. It also reduced the time it took participants to listen to the recognition result and decide if there was an error by 11%.
more »
« less
This content will become publicly available on April 11, 2026
Detecting Music Performance Errors with Transformers
Beginner musicians often struggle to identify specific errors in their performances, such as playing incorrect notes or rhythms. There are two limitations in existing tools for music error detection: (1) Existing approaches rely on automatic alignment; therefore, they are prone to errors caused by small deviations between alignment targets; (2) There is insufficient data to train music error detection models, resulting in over-reliance on heuristics. To address (1), we propose a novel transformer model, Polytune, that takes audio inputs and outputs annotated music scores. This model can be trained end-to-end to implicitly align and compare performance audio with music scores through latent space representations. To address (2), we present a novel data generation technique capable of creating large-scale synthetic music error datasets. Our approach achieves a 64.1% average Error Detection F1 score, improving upon prior work by 40 percentage points across 14 instruments. Additionally, our model can handle multiple instruments compared with existing transcription methods repurposed for music error detection.
more »
« less
- Award ID(s):
- 2326198
- PAR ID:
- 10626351
- Publisher / Repository:
- PKP Publishing Services Network
- Date Published:
- Journal Name:
- Proceedings of the AAAI Conference on Artificial Intelligence
- Volume:
- 39
- Issue:
- 22
- ISSN:
- 2159-5399
- Page Range / eLocation ID:
- 23687 to 23695
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Alkan, Can (Ed.)Abstract MotivationDetection of structural variants (SVs) from the alignment of sample DNA reads to the reference genome is an important problem in understanding human diseases. Long reads that can span repeat regions, along with an accurate alignment of these long reads play an important role in identifying novel SVs. Long-read sequencers, such as nanopore sequencing, can address this problem by providing very long reads but with high error rates, making accurate alignment challenging. Many errors induced by nanopore sequencing have a bias because of the physics of the sequencing process and proper utilization of these error characteristics can play an important role in designing a robust aligner for SV detection problems. In this article, we design and evaluate HQAlign, an aligner for SV detection using nanopore sequenced reads. The key ideas of HQAlign include (i) using base-called nanopore reads along with the nanopore physics to improve alignments for SVs, (ii) incorporating SV-specific changes to the alignment pipeline, and (iii) adapting these into existing state-of-the-art long-read aligner pipeline, minimap2 (v2.24), for efficient alignments. ResultsWe show that HQAlign captures about 4%–6% complementary SVs across different datasets, which are missed by minimap2 alignments while having a standalone performance at par with minimap2 for real nanopore reads data. For the common SV calls between HQAlign and minimap2, HQAlign improves the start and the end breakpoint accuracy by about 10%–50% for SVs across different datasets. Moreover, HQAlign improves the alignment rate to 89.35% from minimap2 85.64% for nanopore reads alignment to recent telomere-to-telomere CHM13 assembly, and it improves to 86.65% from 83.48% for nanopore reads alignment to GRCh37 human genome. Availability and implementationhttps://github.com/joshidhaivat/HQAlign.git.more » « less
-
Spatial sound reasoning is a fundamental human skill, enabling us to navigate and interpret our surroundings based on sound. In this paper we present BAT, which combines the spatial sound perception ability of a binaural acoustic scene analysis model with the natural language reasoning capabilities of a large language model (LLM) to replicate this innate ability. To address the lack of existing datasets of in-the-wild spatial sounds, we synthesized a binaural audio dataset using AudioSet and SoundSpaces 2.0. Next, we developed SpatialSoundQA, a spatial sound-based question-answering dataset, offering a range of QA tasks that train BAT in various aspects of spatial sound perception and reasoning. The acoustic front end encoder of BAT is a novel spatial audio encoder named Spatial Audio Spectrogram Transformer, or Spatial-AST, which by itself achieves strong performance across sound event detection, spatial localization, and distance estimation. By integrating Spatial-AST with LLaMA-2 7B model, BAT transcends standard Sound Event Localization and Detection (SELD) tasks, enabling the model to reason about the relationships between the sounds in its environment. Our experiments demonstrate BAT's superior performance on both spatial sound perception and reasoning, showcasing the immense potential of LLMs in navigating and interpreting complex spatial audio environments.more » « less
-
The robustness and vulnerability of Deep Neural Networks (DNN) are quickly becoming a critical area of interest since these models are in widespread use across real-world applications (i.e., image and audio analysis, recommendation system, natural language analysis, etc.). A DNN's vulnerability is exploited by an adversary to generate data to attack the model; however, the majority of adversarial data generators have focused on image domains with far fewer work on audio domains. More recently, audio analysis models were shown to be vulnerable to adversarial audio examples (e.g., speech command classification, automatic speech recognition, etc.). Thus, one urgent open problem is to detect adversarial audio reliably. In this contribution, we incorporate a separate and yet related DNN technique to detect adversarial audio, namely model quantization. Then we propose an algorithm to detect adversarial audio by using a DNN's quantization error. Specifically, we demonstrate that adversarial audio typically exhibits a larger activation quantization error than benign audio. The quantization error is measured using character error rates. We use the difference in errors to discriminate adversarial audio. Experiments with three the-state-of-the-art audio attack algorithms against the DeepSpeech model show our detection algorithm achieved high accuracy on the Mozilla dataset.more » « less
-
pyAMPACT: A Score-Audio Alignment Toolkit for Performance Data Estimation and Multi-modal ProcessingpyAMPACT (Python-based Automatic Music Performance Analysis and Comparison Toolkit) links symbolic and audio music representations to facilitate score-informed estimation of performance data from audio. It can read a range of symbolic formats and can output note-linked audio descriptors/performance data into MEI-formatted files. pyAMPACT uses score alignment to calculate time-frequency regions of importance for each note in the symbolic representation from which it estimates a range of parameters from the corresponding audio. These include frame-wise and note-level tuning-, dynamics-, and timbre-related performance descriptors, with timing-related information available from the score alignment. Beyond performance data estimation, pyAMPACT also facilitates multi-modal investigations through its infrastructure for linking symbolic representations and annotations to audio.more » « less
