Questions tagged [speech-recognition]
Translation of spoken words into text, usually with the help of computers.
48
questions
-1
votes
1
answer
32
views
Terminologies in Acoustic-phonetic Vowel classifier
An acoustic-phonetic vowel classifier uses a tree based classification.
Consider F1 and F2 as formant frequencies. At it's first step, it checks for COMPACT/DIFFUSE (High F1/Low F1). At the second, it ...
0
votes
0
answers
21
views
Investigating the role of Functional Load in Speech Recognition
I am currently delving into the application of Stokes and Surendran’s Functional Load (FL) in the context of Dutch CVC (Consonant-Vowel-Consonant) words. Stokes and Surendran (2005) propose FL as a ...
- 121
2
votes
0
answers
28
views
Are there any free tools to integrate in apps for evaluating speech on a word level?
I want to give the sound of a word and ask the user to repeat it.
In the app, I want to be able to check the accuracy of the user's speech and give him feedback on whether he said it correctly or not.
...
- 41
0
votes
0
answers
138
views
Which datasets can I use to train a medical speech recognition system?
So far I found:
ezDI Medical Dictation Dataset (16h)
https://www.kaggle.com/datasets/paultimothymooney/medical-speech-transcription-and-intent (8h)
Which other datasets can I use to train a medical ...
- 1,588
0
votes
0
answers
114
views
Why does Whisper ASR sometimes have a really bad output?
The Whisper ASR is a very accurate speech recognition system, on average. But in some rare cases, it has very bad output.
E.g.:
ground truth
whisper output
seven
damn it!
her jewelery shimmered
Hey,...
- 1,588
0
votes
0
answers
44
views
What's the state-of-the-art tool for ASR (automatic speech recognition) Error Detection?
ASR Error Detection = automatically detect erroneous words in ASR output.
The survey {1} was published 4 years ago. {2} didn't release their implementation.
References:
{1} Errattahi, Rahhal, Asmaa ...
- 1,588
0
votes
0
answers
1k
views
What's the state-of-the-art, word-level forced alignment tool that is ok to use commercially?
Forced alignment = synchronize audio and text. I'm looking for word-level forced alignment.
I see many forced alignment tools:
https://github.com/topics/forced-alignment
https://github.com/pettarin/...
- 1,588
3
votes
0
answers
43
views
Speed listening
I have just heard that there are people out there (mostly blind people) using screen readers at overspeed, achieving speech rates twice as high as usual, and even higher.
What I want to know is: Are ...
- 30.2k
2
votes
1
answer
73
views
Up-to-date performance comparison of speech recognition over time
I am looking for an up-to-date (as of 2022) figure or table which displays the improvement in speech recognition on benchmarks for read and spontaneous speech over time, ideally in terms of word error ...
- 508
0
votes
1
answer
144
views
English / French speech to IPA
Are there any open-sourced library / Deep Learning models that convert an audio clip of a word to its IPA representation? In this case, the audio is from a non-native speaker and the goal is to ...
-1
votes
1
answer
56
views
Does a voice activated conversation with an AI constitute as natural speech?
Since AI bots fulfil all the requirements of speech, but don't fulfil the criteria of 'natural speech' production since it wasn't done impromptu and wasn't developed naturally, do you guys think that ...
2
votes
1
answer
83
views
Are there any set guidelines for the lifespan of a (spoken) corpus?
Are there any set guidelines on how long a (spoken) corpus stays relevant / is reliable to use? The only major Dutch spoken corpus is from 2004, and I was wondering if using this corpus would cause a ...
2
votes
2
answers
265
views
Is it possible to automatically recognize a single syllable as its IPA transcription?
Is it possible to take a sound file as the input, in which there is a single syllable, and we can get the IPA transcription of the syllable automatically? If it is, how? Or to say is there already a ...
- 301
0
votes
2
answers
360
views
Distinguishing between [s] and [ʪ] in spectrogram
Is there is a way to differentiate between [s] and [ʪ] using spectral analysis (Pratt or spectrum view or any other software)? Is there is any particular pattern that only appears with the [ʪ]?
[ʪ] ...
- 137
2
votes
1
answer
227
views
How to read and understand linguistics articles?
I was wondering if there is a good way to read and understand Quantitative linguistics articles that has graphs in it?
For example,
For a class, I am currently reading: "Recognition of spoken ...
- 194
2
votes
1
answer
2k
views
What's the relationship between harmonics and formants?
I'm new in speech-processing and there are some notions confused me.
Through Fourier Transform, we can get spectra of specific sound signals, the harmonic frequencies can be shown as the peaks in the ...
- 301
3
votes
0
answers
75
views
Computational linguist's estimate of minimum hours of speech needed to train a neural net to recognize speech
From a computational linguist's point of view, is there a lower limit on the number of hours of speech needed to train a neural net to translate speech to text? An estimate from CMU is 3000-5000 hours ...
- 131
1
vote
1
answer
61
views
How important is speaker-balance for speech recognition models?
Working on ASR models, I have encountered several datasets which have distributions where a small amount of speaker make a huge part of the actual dataset.
The following image shows the extracted ...
- 121
2
votes
2
answers
66
views
How do ASR systems cope with noise?
(Apologies if this question is not specific enough.) I am researching parsing strategies in noisy environments, and how noise changes these strategies. Ultimately I would like to understand how noise ...
- 576
1
vote
1
answer
112
views
What problems does co-articulation cause in speech recognition?
I know it becomes harder to find the start and end of phonemes due to co-articulation. I want to know about other problems that co-articulation causes in speech recognition.
- 81
3
votes
3
answers
172
views
What do you call the same phoneme that behaves a little bit differently in particular environments?
Not Allophones!
My point is, when the same phoneme looks different in different places (has the same IPA representation, but look different on spectrograms) (regarding to speech analysis): (like the ...
- 1,251
1
vote
2
answers
1k
views
Libs or APIs for removing background noise from voice input?
The effect of background noise on speech recognition is not good.
Are there libs or APIs for removing background noise from voice input?
Ideally language-agnostic or tuned for the top 100+ languages,...
- 7,664
0
votes
1
answer
46
views
Finding fundamental particles of sound In ASR systems?
Q1. Can we divide phonemes which are specific to a language into some more fundamental entities same across all languages ?
Q2. Will this help us in achieving language independent acoustic modelling ?...
- 109
2
votes
1
answer
945
views
Turkish voice to text speech recognition references
Does anybody know what is approach implemented in Google Speech API and Siri for Turkish voice to text speech recognition?
I'm interested in details of these two services, not the general or ...
4
votes
2
answers
4k
views
Any software that label a WAV file into phonemes
I have a WAV file contains a subject speech. The subject speaks a sentence once at a time, then a short period of silent appears. I'm interested to analyze the phonemes of that speech and what time ...
- 143
2
votes
1
answer
916
views
looking for corpus of dialogue recording in appointment between doctors and patients
I am looking for a public corpus of dialogue recording in short/medium appointment between doctors and patients, aims to study speech variation, the prosody of speech, measuring speech' rhythm, ...
- 43
1
vote
0
answers
189
views
How does Microsoft speech recognition API compare with Google Cloud Speech API in terms of speech recognition accuracy?
How does Microsoft speech recognition API compare with Google Cloud Speech API in terms of speech recognition accuracy (e.g., in terms of word error rate, character error rate (CER), or sentence error ...
- 1,588
8
votes
2
answers
13k
views
How does Kaldi compare with Mozilla DeepSpeech in terms of speech recognition accuracy?
How does Kaldi ASR compare with Mozilla DeepSpeech in terms of the speech recognition accuracy (e.g., in terms of word error rate)?
- 1,588
3
votes
2
answers
678
views
Why is recognition based on phonemes and not syllables or morphemes?
Why is recognition based on detecting individual phonemes, and not chunks such as morphemes or syllables?..
My question were in relation to human perception, and the existence of phonetic boundary, ...
- 169
1
vote
2
answers
100
views
Acoustic signal to phones?
I am currently trying to understand speech perception, and are at the moment a bit stuck.. I seem to understand how the ear process incoming sound, but i don't understand the concept of phones. Why ...
- 169
2
votes
1
answer
119
views
Homemade ASR decoding
I am looking for a good decoder to pair my acoustic model output with my language model output.
Since my acoustic model outputs a string of phonemes they need to be fuzzy matched with the ...
- 121
0
votes
2
answers
195
views
What phonetic elements is the word yes and no put together by? [closed]
This might be a long shot but it will try it anyway?...
I am currently working on developing a speech recognition application, capable of detecting utterances consisting of the words yes/no.
A ...
- 169
1
vote
0
answers
50
views
Speech Recognition: prosody. What is the difference between combined and intergrated models?
I am readin a paper about dialogue modelling and the identification of dialogue acts using prosody. The paper describes 2, maybe three types of model for utilizing the speech signal: a separate, ...
- 21
4
votes
2
answers
609
views
How to read Spectrograms using Praat Script
I am new to Praat and trying to read the spectrograms of different sound files using these videos. I am learning to read the specrograms using the UI of Praat but I am unable to find any way through ...
- 205
0
votes
1
answer
152
views
How to identify the English \t\ consonant in sound recordings?
I am trying to recognize the consonants in a word i.e. if the word spoken is tap, using different tools like Praat or MATLAB etc. I want to implement a system to ensure that the first letter spoken ...
- 205
2
votes
2
answers
979
views
How to find amplitude values on each time unit in a sound file
I am new to Praat and tryingt to get the selected region in Praat script using the the amplitude values i.e. when amplitude gets larger than a threshold then I should start capturing the region and ...
- 205
2
votes
1
answer
407
views
From where I can get phonemes in audio format?
I am working on a speech recognition system and to begin with I need all phonemes of English in audio format by which I will be able to compare with my speech fragments.
- 121
5
votes
2
answers
828
views
Algorithm for Separating Consonant and Vowel Waveforms from Speech Signal
Is anyone aware of a software package or general algorithm that allows the separation of consonant and vowel waveforms from a continuous speech signal?
I've tried to implement the technique "Harmonic/...
- 51
7
votes
1
answer
346
views
Indistinguishability of [f] and [s] on phone - citation?
I have long felt that [f] and [s] are hard to tell apart on the phone, especially when spelling out words letter by letter. As a non linguist (but audio engineer) it seems to me that the frequencies ...
- 173
2
votes
1
answer
416
views
Why is [f] ambiguous between [f] and [s] after saying the word <three>?
I used an automated customer service system that requires reading off one's case id. The case id I had included the sequence "3F."
The speech recognition software was only tripped up by that "F,"...
- 484
2
votes
1
answer
417
views
What is harder? Speech synthesis or speech recognition?
Supposing to use Festival to build a speech synthesis system and HTK for a speech recognition system.
The language is a Native American language.
Which system would be harder to accomplish?
- 29
3
votes
2
answers
200
views
Are there any good open source speech to text trancription tools or programs?
I am working on a project which needs to convert a given speech (in an audio file,only speech no music and other stuffs) into text and later analyze that text.
It would be a great help if I could able ...
- 41
0
votes
0
answers
258
views
Is there any speech to IPA software? [duplicate]
Does anyone know of a software that does speech to IPA transcription? It seems like this is an easier problem than speech to text but searching for it online didn't give me any results.
- 109
24
votes
7
answers
23k
views
Convert audio recording of word to IPA representation
Are the any open source tools/software libraries to convert an audio clip to its IPA representation? If so, are they accurate? If not, why not?
Here is a Gaelic word I wish to convert:
Ogg format:
...
- 1,072
3
votes
0
answers
82
views
To what extent is aural language comprehension based on a post-processing model, and what are the factors involved?
It occurred to me that what we hear and interpret as speech is often an inarticulate garble of phonemes that native speakers process with lightning speed (usually) to come up with a clear and specific ...
- 802
3
votes
0
answers
2k
views
What does "parsimonious representation" mean in the context of speech recognition?
I was studying about LPC (Linear Predictive coding) and in this topic the author was explaining the reasons why LPC is used widely for speech recognition system. One reason he gave was this:
The ...
- 183
5
votes
3
answers
8k
views
Difference between production of vowels, diphthongs and semi-vowels
I am studying speech recognition by Lawrence Rabiner's book. I am unable to find a proper and easy to understand answer for the following question :
Difference between production of vowels, ...
- 183
4
votes
0
answers
119
views
Determining how similar audio is to human speech
I was suggested to cross post this question of mine from Stack Overflow. Hopefully my question is within the bounds of this question board!
I am searching for a method of determining the similarity ...
- 141