{"id":"52ea2352-84fb-431e-ac58-7c402cf4b8ed","arxiv_id":"2604.25938","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":2.0,"correctness_risk":"high","formal_verification":"none","parameter_count":1,"one_line_summary":"MFCC features with an LSTM classifier reach 99% accuracy on emotion recognition from the TESS speech dataset, marginally above an SVM baseline.","lead":"This paper describes extracting Mel-Frequency Cepstral Coefficients from speech audio and feeding them into an LSTM neural network to classify human emotions on the TESS dataset, reporting 99% accuracy compared to 98% for an SVM baseline. A smart generalist might read it to see how standard deep learning tools can be applied to voice interfaces for tasks like virtual assistants or basic mental health monitoring.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.3","headline":"99% accuracy likely relies on non-speaker-independent split of TESS (only 2 speakers)","rationale":"The reader's weakest assumption matches the load-bearing point exactly. The full manuscript (standard pipeline, no novelty) provides no additional evidence such as speaker-independent results, ablation studies, or external validation sets that would offset the concern. This keeps the verdict at UNVERDICTED with high correctness risk.","tokens_in":1788,"tokens_out":363,"duration_ms":25338,"concrete_test":"Locate the experimental setup section describing data partitioning and model training; if a random 70/30 or 80/20 split is used without speaker separation, re-run the MFCC+LSTM pipeline on TESS using leave-one-speaker-out cross-validation (train on one speaker, test on the other, average); if mean accuracy falls below 85%, the original 99% figure does not substantiate the emotion-pattern claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"TESS contains utterances from just two female speakers. The central claim that the MFCC-LSTM model 'succeeds in capturing the patterns of emotions' and delivers 'highly realistic classifications' requires that accuracy reflect emotion-specific features rather than speaker identity or recording artifacts. The abstract reports a 99% LSTM accuracy versus 98% SVM baseline but gives no description of the train/test split, cross-validation scheme, or speaker-independent protocol. A standard random split would place utterances from the same speaker in both sets, allowing the LSTM to exploit speaker-dependent cues (pitch, timbre) that correlate with the acted emotions in this small dataset. Without explicit leave-one-speaker-out or speaker-disjoint partitioning, the reported accuracy does not support the generalization implied by the claim.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript presents a speech emotion recognition (SER) system that extracts Mel-Frequency Cepstral Coefficients (MFCC) from the Toronto Emotional Speech Set (TESS) dataset and feeds them into an LSTM classifier, reporting 99% accuracy versus 98% for an SVM baseline with RBF kernel. The abstract claims the MFCC-LSTM approach captures emotion patterns and yields highly realistic classifications across the dataset's emotion classes.","tokens_in":1949,"tokens_out":476,"duration_ms":38152,"significance":"If the 99% accuracy were obtained under a speaker-independent protocol with proper cross-validation, the result would provide empirical evidence that LSTM networks can model sequential dependencies in MFCC features for SER on TESS. The inclusion of an SVM baseline offers a minimal but useful point of comparison. However, the current lack of methodological detail on evaluation prevents any assessment of whether the performance reflects genuine emotion modeling or dataset artifacts.","major_comments":[{"comment":"Abstract and Results: The central claim that the LSTM model 'succeeds in capturing the patterns of emotions in speech' and achieves '99% accuracy' is unsupported because the manuscript supplies no information on the train-test split, cross-validation scheme, hyperparameter selection, or error analysis. TESS contains utterances from only two speakers; without an explicit speaker-disjoint (leave-one-speaker-out) partition, the reported accuracy cannot be interpreted as evidence of emotion-specific generalization rather than speaker or recording artifacts.","section":"Abstract and Results"}],"minor_comments":[{"comment":"Abstract: The phrasing 'the author introduces' is inconsistent with standard academic style; 'this work introduces' or 'we introduce' would be preferable. The clause 'the sound similarity between specific feelings' is unclear and should be rephrased for precision.","section":"Abstract"},{"comment":"Overall: No architecture diagram, layer dimensions, or training hyperparameters for the LSTM are provided, making the model non-reproducible from the text alone.","section":"Methods"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is very short and reads like a conference extended abstract; confirm whether it meets the journal's expectations for a full-length article in terms of methodological depth and related-work coverage."},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for their thorough review and constructive criticism. We respond to the major comment as follows and will incorporate the suggested improvements in the revised manuscript.","responses":[{"response":"We acknowledge the validity of this comment. The submitted manuscript indeed omitted key details about the evaluation methodology, which is a significant shortcoming. In the revised version, we will add a detailed 'Experimental Setup' section that specifies the train-test split procedure, the cross-validation scheme employed, the hyperparameter selection process, and an error analysis including confusion matrices. Furthermore, we will explicitly discuss the dataset's composition (only two speakers) and the implications for generalization. We will clarify the nature of the split used and include additional analysis to support the claims about capturing emotion patterns. The abstract will be revised to more accurately reflect the supported claims.","revision_made":"yes","referee_comment":"[Abstract and Results] Abstract and Results: The central claim that the LSTM model 'succeeds in capturing the patterns of emotions in speech' and achieves '99% accuracy' is unsupported because the manuscript supplies no information on the train-test split, cross-validation scheme, hyperparameter selection, or error analysis. TESS contains utterances from only two speakers; without an explicit speaker-disjoint (leave-one-speaker-out) partition, the reported accuracy cannot be interpreted as evidence of emotion-specific generalization rather than speaker or recording artifacts."}],"tokens_in":1411,"tokens_out":306,"duration_ms":31291,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The central issue is that the claimed 99% LSTM accuracy on TESS does not demonstrate real emotion recognition. TESS contains acted speech from just two female speakers. A standard random train-test split would let the model pick up speaker identity or recording artifacts rather than emotion cues, and the abstract supplies no information on splits, cross-validation, or speaker-disjoint testing. That makes the performance number hard to trust for anything beyond this tiny dataset.","headline":"The 99% accuracy is probably not reliable because TESS has only two speakers and the paper gives no sign of a speaker-independent split.","tokens_in":2471,"tokens_out":162,"would_cite":false,"duration_ms":18732,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"MFCC features fed into an LSTM network recognize speech emotions at 99 percent accuracy.","keywords":["speech emotion recognition","MFCC features","LSTM","deep learning","TESS dataset","emotion classification","human-computer interaction"],"falsifier":"Testing the model on an independent speech dataset recorded under different conditions or with new speakers and finding significantly lower accuracy would disprove the claim of effective emotion pattern capture.","tokens_in":2729,"feed_emoji":"🗣️","tokens_out":598,"duration_ms":35934,"temperature":0.7,"pith_summary":"This paper develops a speech emotion recognition system by extracting Mel-Frequency Cepstral Coefficients from audio signals and processing them with a Long Short-Term Memory deep learning model. The system is evaluated on the Toronto Emotional Speech Set, achieving 99 percent accuracy across multiple emotion classes, which exceeds the 98 percent from a support vector machine baseline. A sympathetic reader would care because accurate emotion detection from voice could enable more responsive human-computer interactions and applications in areas like mental health monitoring. The work establishes that LSTM networks can effectively learn the temporal patterns in speech modified by emotional states.","feed_headline":"LSTM with MFCC features reaches 99% accuracy on speech emotions","feed_subtitle":"The model beats an SVM baseline on TESS data and supports uses in virtual assistants and mental health monitoring.","key_machinery":"Mel-Frequency Cepstral Coefficients extracted from speech signals, used as input to a Long Short-Term Memory neural network for classifying emotional states.","core_discovery":"The authors show that transforming pre-processed speech signals from the TESS dataset into MFCC features and inputting them to an LSTM model enables the network to learn long-term sequential dependencies, resulting in highly accurate classification of emotions such as those present in the dataset. This LSTM-based classifier outperforms a classical SVM with RBF kernel, which reaches only 98 percent accuracy. The results confirm that such architectures are suitable for the task of speech emotion recognition.","pith_inferences":["If the accuracy holds on datasets beyond TESS, the system could handle real-world variations in speakers and conditions more reliably.","Combining this audio-based method with visual cues might create more robust multimodal emotion recognition.","Further testing on diverse recording environments would clarify the model's practical limits."],"forward_implications":["The MFCC-LSTM approach captures emotional patterns in speech effectively.","It delivers highly realistic classifications for all selected emotion classes.","The LSTM model can be applied to address speech emotion recognition tasks.","Potential uses include virtual assistants and mental health surveillance systems."],"fun_headline_variants":["LSTM MFCC hits 99% on TESS emotion recognition","LSTM scores 99% accuracy with MFCC for speech emotions","MFCC inputs to LSTM yield 99% emotion accuracy","LSTM based model reaches 99% on TESS speech emotions"],"cache_read_input_tokens":64,"weakest_assumption_plain":"The model's high performance on the TESS dataset means it has learned generalizable features of emotional speech rather than patterns unique to that collection of recordings.","fun_headline_variants_meta":{"raw":{"variants":["LSTM MFCC hits 99% on TESS emotion recognition","LSTM scores 99% accuracy with MFCC for speech emotions","MFCC inputs to LSTM yield 99% emotion accuracy","LSTM based model reaches 99% on TESS speech emotions"]},"model":"grok-4.3","cost_usd":0.010785,"raw_usage":{"total_tokens":4725,"prompt_tokens":768,"num_sources_used":0,"completion_tokens":68,"cost_in_usd_ticks":107853000,"prompt_tokens_details":{"text_tokens":768,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":3889,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":768,"tokens_out":68,"duration_ms":24615,"temperature":1.0,"reasoning_tokens":3889,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-05-10T09:55:06.221898+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Testing the model on an independent speech dataset recorded under different conditions or with new speakers and finding significantly lower accuracy would disprove the claim of effective emotion pattern capture.","supporting_citations":[],"review_version":1}