Skip silence

[chirag127]

Skip silence

[ImprovedTube]

yay! + another conditional speed/skip based on content? (#899 (comment) )

• technically simple implemenation: Dynamic speed based on speed of speech (syllables per minute or wpm) & difficulty (average letters per word) (based on transcript & category - not audio)

• hi @vantezzen https://github.com/vantezzen/skip-silence (MIT license!)

What kinds of silences are typical?

1. phonetics (& necessary physiology). ~<0.7 seconds each. Some of these pauses make words(/phrases) more recognizable or just familiar (even if their origin would be physiology/hard to avoid physically).

• So if some of these could be skipped to save time efficiently, removing others might increase listening-difficulty/listening-errors.

2. category dependend error. I.e. in Live stream & recordings of lectures.
   The less scripted the sound / talk was, the more likely there are pauses due to human error, lack of performance/exhaustion of the speaker, combined with 'um'- & 'hm'-sounds (or other errors, like mono #552).

• if such category is detected, then the remedy can be much more specific/confident.

3. 0.5s-10s rethorics (like minute/s of silence, to express relevance solemnly/diginity/equanimity)

• hard to decide. maybe easier by category or by channel

4. several seconds or minutes: broken sound/error, silent scenes/videos(no sound existing), art

• these can be skipped in inactive tabs more likely meaningfully than in active ones.

• in active, visible tabs, all silence-skipping could(?) analyze video too. ( - Video frames should not change much noticably during the whole skipped period)

• 6. Exclusion/s: Music, .., ... ?

On 1. : Is all existing (proportional) digital speed-up only skipping random parts?
"Micro silences" appearing everywhere, could be removed from every syllable or word in a speech, without changing their proportional tempo. Thus changing how the speaker's energy is perceived, but no changing the clarity of words as much - nor the relational speed (pauses might be semantically relevant. )

• I.e. maybe cutting 10% of the duration of each unit of speech (1x syllable/word+ 1x pause)
  • This might, range from 0.2s to 0.6s usually for one speaker/video (up to 10 seconds rethorically, rarely)
    In this case maybe each could be shortned by 10%. i.e. 0.09s to 0.54s (up to 4.5s).
    • Could be applied to music more obviously, if every beat can be 0.5seconds duration, and might include at least 0.05s in silence/s between the notes played or syllables sung during that beat. ( - Yet the single beat or word might change the rythm wihin)

[chirag127]

phonetics (& necessary physiology). ~<0.7 seconds each. Some of these pauses make words(/phrases) more recognizable or just familiar (even if their origin would be physiology/hard to avoid physically).

We should skip it .

2. category dependent error. I.e. in Live stream & recordings of lectures.

related to

5. Exclusion/s: Music, .., ... ?

• 0.5s-10s rethorics (or minutes of silence, to express relevance solemnly/diginity/equanimity)

We should skip it .

• All silence-skipping could analyze video too

I don't think it is needed. Frame may change a lot during a segment which is intended to be skipped and hence create too many false positive.

• these can be skipped in inactive tabs more likely meaningfully than in active ones.

we should skip it in both active and inactive ones .

5. Exclusion/s: Music, .., ... ?

Of course, if easily possible. May provide opt-out option for disabling skip silence in music .

Micro silences" appearing everywhere,

May provide opt-in option for enabling micro silence skipping or just do it by default and not provide option to opt-out of it .
as I think skip silence is absolute skip silence to save time .

• less complicated feature idea: dynamic speed based on speed of speech (syllables per minute or wpm)

Yeah, it is very easy if there are captions in the video .

• dynamic speed based on speed of speech (syllables per minute or wpm

How will we calculate the WPM in the video that don't have captions ?

[ImprovedTube]

*why would you want to skip when a lot is happening in the video and the tab is active too? (i.e. new scene, completely different backgrounds)

[chirag127]

*why would you want to skip when a lot is happening in the video and the tab is active too? (i.e. new scene, completely different backgrounds)

I just thought it will quite complicated and performance inefficient, So, I had said that.

[allanlaal]

in my experience with listening to 2000h+ of podcasts per year: skipping silence becomes a non-issue and even annoying when you have the proper playback speed set (for me 2.3..3.0x, but there's a legend of a 8.0x user )

[chirag127]

in my experience with listening to 2000h+ of podcasts per year: skipping silence becomes a non-issue and even annoying when you have the proper playback speed set (for me 2.3..3.0x, but there's a legend of a 8.0x user )

By the way, Google podcast also have a option to skip silence but YouTube does not.

[ImprovedTube]

Added skipping silence to the list of player buttons we should add, so that it can just be activated per video or mood. #1445
(While each button on right-click could allow to build specific sticky rules 'always for this channel' / category / ..

---

Hi guys! @allanlaal what about 0.8x while working?
@chirag127 ...performance, yes. Just tried to be complete ☺