Re: [AUDITORY] Cut silence in beginning and end of speech recordings automatically? (Christine Rankovic )

Subject: Re: [AUDITORY] Cut silence in beginning and end of speech recordings automatically? From: Christine Rankovic <rankovic@xxxxxxxx> Date: Fri, 5 Mar 2021 09:35:31 -0500 This is a multi-part message in MIME format. ------=_NextPart_000_0007_01D711A2.E30CCCC0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Hello Tamar: =20 The noise at the beginning and ending of recorded waveforms is annoying. = It is distracting to listeners and the waves should be carefully = trimmed. =20 Edit by hand if at all possible, especially if the waves are intended = for intelligibility testing and consist of nonsense syllables, words, or = sentences. You=E2=80=99ll see that it is easy to inadvertently cut off = low-intensity speech sounds such as =E2=80=98s=E2=80=99, = =E2=80=98th=E2=80=99, etc. Also, make sure to cut the wave only at = zero-crossings to avoid clicks.=20 =20 Wave editing isn=E2=80=99t difficult if you practice on a set of waves = and listen carefully to the results. I can=E2=80=99t imagine that = automatic editing methods can achieve this. =20 =20 Best wishes,=20 Christine Rankovic, PhD =20 =20 From: AUDITORY - Research in Auditory Perception = [mailto:AUDITORY@xxxxxxxx On Behalf Of Gabriele Bunkheila Sent: Friday, March 05, 2021 5:20 AM To: AUDITORY@xxxxxxxx Subject: Re: Cut silence in beginning and end of speech recordings = automatically? =20 Hi Tamar, =20 Since you mentioned MATLAB, I thought I=E2=80=99d share a couple of = pointers. A good fit for this would be detectSpeech = (https://www.mathworks.com/help/audio/ref/detectspeech.html), which uses = a fairly accessible algorithm based on short-term energy and spectral = spread. detectSpeech has been available in Audio Toolbox since release = R2020a. =20 In case any of your data was more challenging, you could consider trying = the function classifySound = (https://www.mathworks.com/help/audio/ref/classifysound.html), which has = only been available since release R2020b and uses the pre-trained YAMNet = network under the hood. =20 I hope this helps =E2=80=93 feel free to get in touch directly if you = needed more guidance.=20 Regards and good luck, Gabriele. =20 -- Gabriele Bunkheila [he/him] =E2=80=93 Product Management, DSP and Audio=20 MathWorks =20 From: AUDITORY - Research in Auditory Perception = <AUDITORY@xxxxxxxx> On Behalf Of Tamar Regev Sent: mi=C3=A9rcoles, 3 de marzo de 2021 16:10 To: AUDITORY@xxxxxxxx Subject: [AUDITORY] Cut silence in beginning and end of speech = recordings automatically? =20 Hi all, =20 Does anyone know of a good way to automatically trim silent parts (which = may contain some minor background noise) at the beginning and end of = speech recordings? =20 Preferentially using Matlab but any other automatic way would work (we = want to run this on many sound files). =20 Thanks a lot! Tamar =20 ------=_NextPart_000_0007_01D711A2.E30CCCC0 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable <html xmlns:v=3D"urn:schemas-microsoft-com:vml" = xmlns:o=3D"urn:schemas-microsoft-com:office:office" = xmlns:w=3D"urn:schemas-microsoft-com:office:word" = xmlns:m=3D"http://schemas.microsoft.com/office/2004/12/omml" = xmlns=3D"http://www.w3.org/TR/REC-html40"><head><meta = http-equiv=3DContent-Type content=3D"text/html; charset=3Dutf-8"><meta = name=3DGenerator content=3D"Microsoft Word 12 (filtered = medium)"><style></style></head><body lang=3DEN-US = link=3D"#0563C1" vlink=3D"#954F72"><div class=3DWordSection1>Hello = Tamar:<o:p></o:p><o:p> </o:p>The noise at the = beginning and ending of recorded waveforms is annoying.=C2=A0 It is = distracting to listeners and the waves should be carefully = trimmed.<o:p></o:p><o:p> </o:p>Edit by hand if at all = possible, especially if the waves are intended for intelligibility = testing and consist of nonsense syllables, words, or sentences.=C2=A0 = You=E2=80=99ll see that it is easy to inadvertently cut off = low-intensity speech sounds such as =E2=80=98s=E2=80=99, = =E2=80=98th=E2=80=99, etc.=C2=A0 Also, make sure to cut the wave only at = zero-crossings to avoid clicks. <o:p></o:p><o:p> </o:p>Wave editing = isn=E2=80=99t difficult if you practice on a set of waves and listen = carefully to the results. =C2=A0I can=E2=80=99t imagine that automatic = editing methods can achieve this.=C2=A0 <o:p></o:p><o:p> </o:p>Best wishes, = <o:p></o:p>Christine Rankovic, PhD<o:p></o:p><o:p> </o:p><o:p> </o:p><div><div = style=3D'border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt 0in = 0in 0in'>From:= = AUDITORY - Research in Auditory Perception = [mailto:AUDITORY@xxxxxxxx On Behalf Of Gabriele = Bunkheila Sent: Friday, March 05, 2021 5:20 AM To: = AUDITORY@xxxxxxxx Subject: Re: Cut silence in beginning = and end of speech recordings = automatically?<o:p></o:p></div></div><o:p> </o:p>Hi = Tamar,<o:p></o:p><o:p> </o:p>Since you mentioned MATLAB, I thought I=E2=80=99d = share a couple of pointers. A good fit for this would be detectSpeech = (<a = href=3D"https://www.mathworks.com/help/audio/ref/detectspeech.html">https= ://www.mathworks.com/help/audio/ref/detectspeech.html</a>), which uses a = fairly accessible algorithm based on short-term energy and spectral = spread. detectSpeech has been available in Audio Toolbox since release = R2020a.<o:p></o:p><o:p> </o:p>In case any of your data was more challenging, you = could consider trying the function classifySound (<a = href=3D"https://www.mathworks.com/help/audio/ref/classifysound.html">http= s://www.mathworks.com/help/audio/ref/classifysound.html</a>), which has = only been available since release R2020b and uses the pre-trained YAMNet = network under the hood.<o:p></o:p><o:p> </o:p>I hope this = helps =E2=80=93 feel free to get in touch directly if you needed more = guidance. Regards and good luck, Gabriele.<o:p></o:p> <o:p></o:p>--<o:p></o:p>Gabriele = Bunkheila [he/him] =E2=80=93 Product Management, DSP and Audio = <o:p></o:p>MathWorks<o:p></o:p><o:p> </o:p><div = style=3D'border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0in = 0in 0in'>From: AUDITORY - Research in = Auditory Perception <AUDITORY@xxxxxxxx> On Behalf Of = Tamar Regev Sent: mi=C3=A9rcoles, 3 de marzo de 2021 = 16:10 To: AUDITORY@xxxxxxxx Subject: = [AUDITORY] Cut silence in beginning and end of speech recordings = automatically?<o:p></o:p></div><o:p> </o:p><div>Hi all,<o:p></o:p><div><o:p> </o:p></div><div>Does anyone know of a good way to = automatically trim silent parts (which may contain some minor = background noise) at the beginning and end of speech = recordings?<o:p></o:p></div><div><o:p> </o:p></div><div>Preferentially using Matlab but any = other automatic way would work (we want to run this on many sound = files).<o:p></o:p></div><div><o:p> </o:p></div><div>Thanks a = lot!<o:p></o:p></div><div>Tamar<o:p></o:p></div><div><o:p> </o:p></div></div></div></body></html> ------=_NextPart_000_0007_01D711A2.E30CCCC0--

This message came from the mail archive
src/postings/2021/
maintained by:

DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University