Translatotron: An End-to-End Speech-to-Speech Translation Model

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by bowmessage on Friday May 17, 2019 @02:00AM (#19936692)

Is it just me or are the final results (Translatotron translation) not playable, while the initial audio samples work fine?

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by barrystaes on Friday May 17, 2019 @02:45AM (#19936903)

Not just you, same problem here. (Firefox on Windows)

Workaround: If you look at the WAV file the player loads (right click > Inspect Element), that URL is playable in a new tab/browser window.

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by fartcannon on Friday May 17, 2019 @03:12AM (#19937073)

Did you by bchance notice what is different regarding the other files?

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by gattilorenz on Friday May 17, 2019 @04:30AM (#19937475)

The original (source) is a PCM S16 LE mono file, 8kHz/16 bit. The translations are 32 bits float LE mono file, 24kHz/32 bit.

Firefox supports 8-bit and 16-bit PCM only ( https://support.mozilla.org/en-US/kb/html5-audio-and-video-f... ). I don't know why, exactly.

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by londons_explore on Friday May 17, 2019 @04:35AM (#19937506)

Thats silly... 16 bit audio really doesn't have enough dynamic range for many uses now. Listen to a bit of classical music with very loud and very quiet bits and you'll hear quantization hiss in the quiet bits.

16 bits allows a dynamic range of 96dB while sounds you might come across in everyday life can peak over 130dB briefly.

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by bildung on Friday May 17, 2019 @12:27PM (#19941085)

While the actual analog sounds may indeed have a higher signal to noise ratio than 96db, most ADCs haven't, i.e. for most digital recordings 16bit should be more than enough.

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by jbay808 on Friday May 17, 2019 @02:10PM (#19942087)

Since audio is usually sampled around 44 kHz, which is slow by signal processing standards, most ADCs can exceed 16 bits by supersampling and averaging the results.

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by a_imho on Friday May 17, 2019 @02:56AM (#19936974)

Same FF on Ubuntu.

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by MrTrvp on Friday May 17, 2019 @02:47AM (#19936917)

Works in chrome

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by gpvos on Friday May 17, 2019 @03:48AM (#19937289)

:angry face:

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by solarkraft on Friday May 17, 2019 @03:06AM (#19937028)

The one official web runtime environment.

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by mcemilg on Friday May 17, 2019 @02:46AM (#19936912)

Same.

[ Parent | on HN ]

Re:Translatotron: An End-to-End Speech-to-Speech Translation... by jonathankchang on Thursday May 30, 2019 @01:27PM (#20053685)

Hey mcemilg,

You asked about NREC earlier but I failed to set up email alerts correctly (and that post is locked now) so I'm replying here if that's ok.

Hiring process takes around a month or two, depending on how schedules line up. There would be a basic phone/skype screen, then a in-depth technical screen on your areas of expertise, and finally an on-site interview. We try to get back to you on each stage within a week of each stage.

We have a bunch of employees that are on H1-B's right now. Let me know if you have any other questions and sorry for the late reply!

[ Parent | on HN ]