VP21/22: TEXT to SPEECH BUG: scrambled imported audio on long text

bitman wrote on 8/9/2024, 5:57 AM

VP22 b93 (and VP21 b315) generated speech is completely garbled/scrambled and a totally unusable noise file after import on the timeline when the text is rather long with a lot of line breaks.

The speech generation itself sounds OK, but after the import the generated audio file is broken.

How to reproduce:

1) select Dutch, Belgium Dena (probably does not matter)

2) copy/past the text into text to speech window

3) import on the timeline

Below is the text to reproduce the bug:

Kluwenklokje (campanula glomerata)
Blaassilene (Silene vulgaris)
Duizendblad (Achillea millefolium)
Rond wintergroen (Pyrola rotundifolia)
Gevlekte orchis (Dactylorhiza maculata)
Bolrapunzel (Phyteuma orbiculare)
Bergcentaurie (Centaurea montana)
Donkere akelei (Aquilegia atrata)
Lelietje-van-dalen (convallaria majalis)
Steenbraam (Rubus saxatilis)
Akeleiruit (Thalictrum aquilegiifolium)
Gele dovenetel (Lamiastrum galeobdolon)
Weideschorpioenvlieg (Panorpa vulgaris)
Bosooievaarsbek (Geranium sylvaticum)
Gewone engelwortel (Angelica sylvestris)
Gewone wijngaardslak (Helix pomatia)
Dagkoekoeksbloem (Silene dioica)
Welriekende salomonszegel (Polygonatum odoratum)
Vogelnestje (Neottia nidus-avis)
Groot hoefblad (Petasites hybridus)
Alpenbergvlas (Thesium alpinum)
Knikkend nagelkruid (Geum rivale)
Geelhartje (Linum catharticum)
Kopjesbekermos (Cladonia fimbriata)
Brilkruid (Biscutella laevigata)
Geel zonneroosje (Helianthemum nummularium)
Kruiptijm (Thymus praecox subsp. articus)
Gewone rolklaver (Lotus corniculatus)
Heksenmelk (Euphorbia esula)
Bloedrode Bremraap (Orobanche gracilis)
Alpen wondklaver (Anthyllis vulneraria subsp. alpicola)
Fraaie vrouwenmantel (Alchemilla mollis)
Veldsalie (Salvia pratensis)
Ribzaad (chaerophyllum hirsutum)
Boszwartkoren (Melampyrum sysvaticum)
Rood dooiermos (Rusavskia elegans)

Comments

bitman wrote on 8/9/2024, 6:02 AM

Note that for example halving the above text (say only use the first 18 lines out of 36 lines) and the import is OK.

Robert Johnston wrote on 8/9/2024, 2:47 PM

@bitman, I don't have Vegas 22, but I do have Vegas 21 (315). Was able to produce audio without problems using the entire list. There was no other media in the project, the project was new. I also translated to English without problems. I chose the "Preview" option in Text to Speech.

bitman wrote on 8/9/2024, 3:02 PM

@Robert Johnston, Did you import?

Because the preview was always OK for me, not the imported audio file.

Robert Johnston wrote on 8/9/2024, 4:01 PM

@bitman, Yes, after I clicked "Generate Preview" and waited, I then clicked "Insert on Timeline." I also tried generating a full preview followed by Insert On Timeline. That audio file I uploaded in my previous post is from when I clicked "Insert on Timeline" -- (contains both Dutch (Belgium) Gena audio and the translation to English audio). I never did play back the 8 second preview to hear how it sounded. I just went straight to "Insert on Timeline." Did you have any speed or pitch adjustments?

From what app did you copy and paste your text into the Text to Speech window? Try copying and pasting the text that you posted online here. Maybe for formats are different. Or did you already try that?

bitman wrote on 8/10/2024, 3:35 AM

@Robert Johnston I redid everything, now using the text copy past from this post, but still the same garbled import file. What I noticed tough is that the imported file has 'weird' media properties (like MPEG Audio):

General
Name: AudioClip20240810-1.wav
Folder: C:\Users\(username)\OneDrive\Documents\VEGAS\VEGAS Projects\VP22\Oostenrijk
Type: MainConcept MPEG-1
Size: 2,65 MB (2 711 040 bytes)
Created: Saturday, August 10, 2024, 10:22:19 AM
Modified: Saturday, August 10, 2024, 10:22:19 AM
Accessed: Saturday, August 10, 2024, 10:22:19 AM
Attributes: Archive

Media information
Stream format: MPEG Audio
Audio stream #1
Audio format: MPEG Audio
Sampling rate: 24000 Hz
Channels: 1 channel
Bit rate mode: Constant
Bit rate: 160000 bps

Streams
Audio: 00:02:15,552, 24 000 Hz; Mono, MPEG

ACID information
ACID chunk: no
Stretch chunk: no
Stretch list: no
Stretch info2: no
Beat markers: no
Detected beats: no

Other metadata
Regions/markers: no
Command markers: no

Media manager
Media tags: no

Plug-In
Name: mcplug2.dll
Folder: C:\Program Files\VEGAS\VEGAS Pro 22.0\FileIO Plug-Ins\mcplug2
Format: MainConcept MPEG-1
Version: Version 22.0 (Build 93)
Company: MAGIX Computer Products Intl. Co.

bitman wrote on 8/10/2024, 3:44 AM

a shorter text with fewer lines produces a good import audio with a different (proper) media type:

Audio format: PCM

General
Name: AudioClip20240810-2.wav
Folder: C:\Users\(username)\OneDrive\Documents\VEGAS\VEGAS Projects\VP22\Oostenrijk
Type: Wave (Microsoft)
Size: 658,64 KB (674 444 bytes)
Created: Saturday, August 10, 2024, 10:36:37 AM
Modified: Saturday, August 10, 2024, 10:36:37 AM
Accessed: Saturday, August 10, 2024, 10:36:37 AM
Attributes: Archive

Media information
Stream format: Wave
Audio stream #1
Audio format: PCM
Sampling rate: 48000 Hz
Channels: 1 channel
Bit rate mode: Constant
Bit rate: 768000 bps

Streams
Audio: 00:00:07,025, 48 000 Hz; 16 Bit; Mono, Uncompressed

ACID information
ACID chunk: no
Stretch chunk: no
Stretch list: no
Stretch info2: no
Beat markers: no
Detected beats: no

Other metadata
Regions/markers: no
Command markers: no

Media manager
Media tags: no

Plug-In
Name: wavplug.dll
Folder: C:\Program Files\VEGAS\VEGAS Pro 22.0\FileIO Plug-Ins\wavplug
Format: Wave (Microsoft)
Version: Version 22.0 (Build 93)
Company: MAGIX Computer Products Intl. Co.

bitman wrote on 8/31/2024, 3:18 AM

For those interested in the actual unplayable (in Vegas) audiofile generated by text to speech:

Here is a link to my dropbox with the file which is unplayable in Vegas 22 timeline (screeching garbish), but playable when played by an external application.

https://www.dropbox.com/scl/fi/xcw8rwnrqbn5uzozqkz93/AudioClip20240810-1.wav?rlkey=cyhyja0sgp6cfo7ozud8eqgte&dl=0