本頁閱讀時間 9 分鐘
概述
PESQ 是 2001 年推出的 ITU-T 標準,能在平均意見分數的量表上預測聽眾對於音質受損錄音的聽感好壞;而 2011 年發表的 STOI 則是用來預測語音的清晰度。這兩者都會將處理過的訊號與乾淨的參考訊號進行比較,且兩者至今仍是語音增強研究中的標準工具。
重點摘要
- PESQ 在類似平均意見分數(mean-opinion-score)的量表上預測感知的語音品質,而 STOI 則將語音清晰度預測為平均相關性,數值越高代表越容易理解。
- PESQ 和 STOI 都是侵入式指標,需要一個與處理後語音對齊的乾淨參考錄音。
- ITU-T P.862 (PESQ) 於 2001 年 2 月核准,並於 2024 年 1 月 5 日被 ITU 列為已刪除,由 P.863 取代。
- STOI 會在約 384 ms 的時間窗內,計算 15 個三分之一倍頻帶的短時包絡線相關性,並將結果進行平均。
- PESQ 的分數會因版本和實作方式而有所不同,因此論文中應指明其使用的版本、模式與程式碼。
深入探討
PESQ (Perceptual Evaluation of Speech Quality) and STOI (Short-Time Objective Intelligibility) are algorithms that estimate how humans would judge a speech recording, without running a listening test. Both are intrusive, also called full-reference, metrics: they need the original clean recording and the processed or degraded version, and they score the second against the first. They answer different questions. PESQ estimates quality, roughly how pleasant and natural the speech sounds. STOI estimates intelligibility, how many words a listener would understand. A noise suppressor can raise one while lowering the other, which is why papers usually report both.
PESQ was standardized by the International Telecommunication Union as Recommendation ITU-T P.862, approved on 23 February 2001 for narrow-band telephone networks and speech codecs. Companion documents followed: P.862.1 in November 2003 defines a mapping from PESQ's raw output to MOS-LQO, a listening-quality score on the same 1 to 5 mean opinion score (MOS) scale used in human tests, and P.862.2 in 2007 extends PESQ to wideband speech sampled at 16 kHz, covering 50 to 7,000 Hz according to Torcoli and colleagues. The ITU's catalog now lists P.862 and its parts as deleted on 5 January 2024 and points users to P.863, titled Perceptual objective listening quality prediction, whose current version dates from March 2018. Despite that, a 2025 paper by Torcoli, Halimeh and Habets notes that PESQ is still widely used.
PESQ works by modeling a listener. The standard pipeline aligns the overall level of both signals, applies a filter that imitates a telephone handset, estimates the time delay between them so the right moments are compared, and splits both into short frames. Kim, El-Khamy and Lee describe the next stages in their 2019 analysis: each frame's spectrum is warped onto the Bark scale, a frequency scale that matches how the ear groups pitches, using 49 bands; the clean and degraded spectra are equalized; and power is converted to perceived loudness with Zwicker's law. The difference in loudness between clean and degraded frames becomes a disturbance. PESQ keeps two versions, a symmetric disturbance and an asymmetric one that penalizes added sounds such as noise more than missing ones. These are averaged over frames and combined into a single score.
STOI was introduced by Taal, Hendriks, Heusdens and Jensen in IEEE Transactions on Audio, Speech, and Language Processing in 2011. It was built for speech that has been processed by time-frequency weighting, the family of methods, such as spectral masks, that turn individual frequency bands up or down over time. As summarized by Kolbaek, Tan and Jensen, STOI resamples both signals to 10 kHz, cuts them into 256-sample frames (25.6 ms) with 50% overlap, and groups the spectrum into 15 one-third octave bands from 150 Hz to about 3.8 kHz. For each band it takes the amplitude envelope over windows of 30 frames, about 384 ms, and computes the correlation between clean and processed envelopes. The final score is the average of these correlations across bands and windows. Because it is an average of correlations, the score can range from -1 to 1, and higher means more intelligible.
The two metrics complement older and newer alternatives. Simple signal measures such as signal-to-noise ratio contain no model of hearing at all. Human listening tests remain the reference but are slow and costly. POLQA, defined in P.863, is the ITU's named successor to PESQ, while PESQ keeps the advantage of several free, open implementations. For intelligibility, Jensen and Taal published extended STOI (ESTOI) in 2016 to handle modulated noise maskers, and the pystoi package offers both. Non-intrusive metrics such as DNSMOS, proposed by Microsoft researchers in 2020, estimate quality without a clean reference, which matters for real recordings where none exists.
The metrics have well-documented weaknesses. Because they are intrusive, they cannot score live calls or field recordings. PESQ was designed for telephone-band speech and handset listening, so its use on modern full-band audio, music or very low bit-rate neural codecs goes beyond what it was validated for. Version confusion is a practical hazard: Torcoli and colleagues found that on one test set, scores from the original P.862 and the wideband P.862.2 differed by a root-mean-square error of 0.62 and by as much as 1.82 points on a single item, and that a 2018 correction to the wideband filter changed scores by an RMSE of 0.56. Their advice is to always state the exact PESQ version and implementation. STOI, for its part, measures envelope correlation rather than understanding, so a higher score is a prediction of better intelligibility, not proof of it.
Optimizing directly for these numbers creates its own risk. Kim and colleagues showed that a differentiable PESQ-style loss improved PESQ and SDR together, but any metric becomes less trustworthy once it is a training target. Good practice pairs PESQ and STOI with at least one other measure, ideally a small human listening test.
The practical requirements are light. Both metrics need time-aligned clean and processed files at a supported sample rate: the pesq package accepts 8 kHz for narrowband mode and 16 kHz for wideband mode, and pystoi takes any rate and resamples internally to 10 kHz. The pesq Python package is MIT-licensed, and pystoi's tests match the original MATLAB implementation within a tolerance of 0.001.
04工作範例
在 Python 中使用 PESQ 和 STOI 為降噪器評分
使用 pip install pesq pystoi soundfile 安裝這些開源套件。
載入一段乾淨的參考錄音,以及加入雜訊後的相同語句,兩者皆為長度相等的 16 kHz 單聲道陣列。
計算基準分數:pesq(16000, clean, noisy, 'wb') 與 stoi(clean, noisy, 16000, extended=False)。
對雜訊檔案執行降噪器,並以相同的取樣率與長度載入其輸出結果。
計算降噪輸出相對於乾淨參考訊號的相同兩項分數。
記錄套件名稱與版本、寬頻模式以及取樣率,並與數值一併保存。
它顯示了什麼
學習者可以獲得品質與清晰度在處理前後的數值。一個有用的降噪器應該能提升 PESQ,並使 STOI 維持不變或更高。作為健全性檢查,將乾淨檔案與其自身進行評分,應會得到約 1 的 STOI 分數,以及該實作所能產生的最高 PESQ 分數。
現實世界的實施
語音增強論文通常會在公開測試集上報告 PESQ 與 STOI。例如,Kim、El-Khamy 與 Lee 在 VoiceBank-DEMAND 與 QUT-NOISE-TIMIT 語料庫上,使用 SDR、PESQ 與 STOI 評估了他們 2019 年的端對端降噪器。
電話與編解碼器工程是 PESQ 的原始任務:2001 年 2 月核准的 ITU-T 建議書 P.862,旨在針對窄頻電話網路與語音編解碼器的端對端品質。
研究人員使用開源的 pesq 套件在 Python 中計算分數,該套件支援 8 kHz 的窄頻評分與 16 kHz 的寬頻評分;此外還有 pystoi 套件,它實作了經典 STOI 與擴展版 STOI。
Kim 與同事們建立了 PESQ 的可微分近似版本,移除了其 IIR 濾波器、延遲搜尋與不良區間疊代,因此可以訓練降噪網路直接提升 PESQ,而不只是降低均方誤差。
Microsoft 研究人員於 2020 年提出了 DNSMOS 作為降噪器的非侵入式品質指標,因為諸如 PESQ 等基於參考的評分需要實際通話中沒有的乾淨錄音。
PESQ 和 STOI 語音品質指標的未來
The field is moving in three directions at once. First, standards have moved on: the ITU lists PESQ as deleted since 5 January 2024 and points to P.863, whose current text dates from March 2018. Second, learned non-intrusive predictors such as Microsoft's DNSMOS (2020) have been proposed that score quality without a clean reference and so can run on real calls. Third, PESQ and STOI remain the common currency of benchmark tables, which keeps them in use for comparison with earlier work. The main open problem is trust. Torcoli, Halimeh and Habets reported in 2025 that differences between PESQ versions are large enough to change conclusions, and they released an implementation of the latest correction, Corrigendum 2, which they say no other open distribution included. Clear reporting of version, sample rate and implementation is becoming an expected part of papers. Another open question is how well these older metrics judge modern generative systems, such as neural codecs and speech generators, that can produce clean-sounding speech which differs from the reference waveform. Learned metrics and small human tests may carry more weight for those systems, though no single replacement has been standardized.
付諸實踐
為什麼這很重要
- 交通與覆蓋範圍. 它透過轉錄、旁白和語音介面提高了可訪問性。
- 成本與預算. 媒體團隊可以用更少的預算更快地交付精美的音訊。
- 速度與規模. 面向客戶的系統可以處理更大規模的語音互動。
留意
- 如果未徵得同意,語音濫用和冒充風險就會增加。
- 由於口音、方言或嘈雜的環境,準確性可能會下降。
- 如果沒有明確的標籤,合成音訊可能會被誤認為是真實的語音。
後續步驟
- 獲得語音捕獲、克隆和重用的明確同意。
- 測試不同揚聲器和背景條件下的品質。
- 定義人員必須審查或批准輸出的時間。
- 標記合成音訊並保留來源記錄以供問責。
資料來源與延伸閱讀
- ITU-TP.862:語音品質感知評估 (PESQ)
- ITU-TP.863:感知客觀聆聽品質預測
- arXiv (ITG Speech Communication 2025)駕馭 PESQ:最新版本與開源實作
- arXiv (IEEE/ACM TASLP)關於語音增強的短時客觀清晰度與短時頻譜振幅均方誤差之間的關係
- arXiv用於感知語音指標聯合優化的端到端多任務降噪
- GitHubpystoi:STOI 的 Python 實作
- GitHubPESQ:適用於 Python 的語音品質感知評估包裝器
- arXivDNSMOS:用於評估降噪器的非侵入式感知客觀語音品質指標
不斷探索
常見問題
什麼是 PESQ 和 STOI 語音品質指標?
PESQ 與 STOI 是用來評判處理後語音最常見的兩個自動化評分。PESQ 是 2001 年的 ITU-T 標準,以平均意見分數尺度預測劣化錄音在聽眾聽來有多好;而於 2011 年發布的 STOI 則預測其清晰度。兩者皆將處理後的訊號與乾淨的參考訊號進行比較,且兩者皆為語音增強研究中的標準工具。
當將劣化語音訊號與參考訊號進行比較時,PESQ 估算的是什麼?
PESQ 旨在作為特定電話和編解碼器條件下感知語音品質的客觀估算;它不會轉錄單字。
STOI 主要預測什麼?
STOI 以 0 到 1 的等級評估清晰度,即語音的可理解程度。
PESQ 和 STOI 都被描述為「侵入性」指標。這意味著什麼?
侵入性指標將降級訊號與乾淨的參考訊號進行比較以計算分數。
STOI 常回報的輸出範圍是多少?
STOI 通常回報在 0 到 1 的尺度上,在評估設定下,較高的分數表示估算的清晰度較高。
PESQ 使用哪一種感知頻率尺度來模擬人類聽力?
PESQ 使用 Bark 頻帶處理將訊號對應到心理聲學響度表示上。