landscape2soundscape.py

import numpy as np
import subprocess
import cv2

# with subprocess and an extra argument 'scene' and a 'resized image saved as png' we can call the server
# yt-dlp is instaled in .d4
# Download Part of Video
# yt-dlp https://www.youtube.com/watch?v=UZ9uyQI3pF0 --downloader ffmpeg --downloader-args "ffmpeg_i:-ss 997 -to 2512"
# ffmpeg -i Sandra\ Kotevska\,\ Painting\ Rose\ bush\,\ mixed\ media\,\ 2017.\ \[NMzC_036MtE\].mkv -f mp3 -ar 22050 -vn out44.wav -ac 1
# https://superuser.com/questions/583393/how-to-extract-subtitle-from-video-using-ffmpeg
#___________________________________________________________________________________________________
#   VIDEO FROM IMAGE with CAPTIONS
#
# UPLOAD to: Simaviro: Documents General WORK PACKAGES WP1 ContentRepository ANBPR_ROMANIA TTSvideos
# __________________________________________________________________________________________________
# TO DONLOAD SRT for youtub
# yt-dlp --write-sub --sub-lang en --convert-subs "srt" https://www.youtube.com/watch?v=F1Ib7TAu7eg&list=PL4x2B6LSwFewdDvRnUTpBM7jkmpwouhPv&index=2
# _voice = 'en_US/vctk_low#p330'
# _voice = 'en_US/cmu-arctic_low#lnh' #en_US/vctk_low#p249'  # 'en_US/vctk_low#p282'
# _voice = ''en_US/vctk_low#p351''
# _voice = 'en_US/vctk_low#p351'  # avoid 318 it does the ghhhhhh
# _voice = 'en_US/m-ailabs_low#judy_bieber'  # Nice voice for ('Arta culinara romaneasca - Groza Irina [phIF0NxgwlQ].mkv' 'Arta culinara romaneasca - Groza Irina [phIF0NxgwlQ].en-GB.srt'),
# _voice = 'en_UK/apope_low'
# _voice = 'en_US/m-ailabs_low#mary_ann'
# _voice = 'en_US/vctk_low#p351'
# _voice = 'en_US/hifi-tts_low#92'
# voice_str = f'_{_voice.replace("/", "")}'





# image/descriptions provided by other SHIFT tool or Human curator

# https://simaviro.sharepoint.com/sites/SHIFT/Shared%20Documents/Forms/AllItems.aspx?csf=1&web=1&e=JNK8dQ&cid=363c253d%2D4d61%2D4db1%2D8ffd%2Ddedda749da2d&RootFolder=%2Fsites%2FSHIFT%2FShared%20Documents%2FGENERAL%2FWORK%20PACKAGES%2FWP1%2FContent%20Repository%2Fshift%5FSPK%5Fuse%5Fcases%5Fshare%2F02%5Fuc%5Fspk%5FLandscape2Soundscape%2FLandscape2Soundscape%5F12%5FMasterpieces&FolderCTID=0x01200058F5037C0101524B82F6F0788C02A563
# STATIC_FRAME = 'uc_spk_Landscape2Soundscape_Masterpieces_pics/01_Schick_AII840_001.jpg' #'assets/image_from_T31.jpg'




PIC_DIR = 'uc_spk_Landscape2Soundscape_Masterpieces_pics/'

DESCRIPTIONS = [
    # 1
    [
        '01_Schick_AII840_001.jpg',                               # image
        '01_Schick_AII840_001.txt',                               # text
        'statue in shire, hill river, vogels.',                  # audiocraft
        'G. Schick, Bildnis der Heinrike Dannecker, 1802', # cv2 puttext title
        'en_UK/apope_low', #'en_US/m-ailabs_low#judy_bieber', #'en_US/m-ailabs_low#mary_ann',
     ],  
    # 2
    [
        '02_Constable_AI555_001.jpg',
        '02_Constable_AI555_001.txt',
        'Meadows country farm village in sight',
        'J. Constable, Dorf An Dem Flusse Stour, 1804',
        'en_US/m-ailabs_low#elliot_miller',
    ],
    # 3
    [
        '03_Schinkel_WS200-002.jpg',
        '03_Schinkel_WS200-002.txt',
        'Arriving at the valley on galloping horses',
        'K. Schinkel Gotische Kirche Auf Einem Felsen 1815',
        'en_US/hifi-tts_low#6097',
    ],
    #
    [
        '04_Friedrich_FV317_001.jpg',
        '04_Friedrich_FV317_001.txt',
        'Land steppes',
        'C. D. Friedrich, Der Watzmann, 1824',
        'en_US/m-ailabs_low#mary_ann',
    ],
    #
    [
        '05_Blechen_FV40_001.jpg',
        '05_Blechen_FV40_001.txt',
        'fjords',
        'C. Blechen, Unwetter In Der Romischen Campagna, 1829',
        'en_US/m-ailabs_low#mary_ann',
    ],
    # 6
    [
        '06_Menzel_AI900_001.jpg',
        '06_Menzel_AI900_001.txt',
        'Olive scenery roman Colloseum',
        'A. Menzel, Bauplatz mit Weiden, 1846',
        'en_US/cmu-arctic_low#aup',
    ],
    # 7
    [
        '07_Courbet_AI967_001.jpg',
        '07_Courbet_AI967_001.txt',
        'Storm at the strand of waves Tsunami',
        'G. Courbet, Die Welle, 1870',
        'en_US/m-ailabs_low#mary_ann',
    ],
    # 8
    [
        '08_Monet_AI1013_001.jpg',
        '08_Monet_AI1013_001.txt',
        'Mai flowers blossom picnic',
        'C. Monet, Sommertag, 1874',
        'en_US/cmu-arctic_low#axb',
    ],
    # 9
    [
        '09_Blechen_AII823_001.jpg',
        '09_Blechen_AII823_001.txt',
        'Cascade in Africa',
        'C. Blechen, Wasserfalle Bei Tivoli, 1832',
        'en_US/m-ailabs_low#mary_ann',
    ],
    # 10
    [
        '10_Boecklin_967648_NG2-80_001_rsz.jpg',
        '10_Boecklin_967648_NG2-80_001.txt',
        'Hades ades at it sisland',
        'A. Bocklin, Toteninsel, 1883',
        'en_US/cmu-arctic_low#jmk',
    ],
    # 11
    [
        '11_Liebermann_NG4-94_001.jpg',
        '11_Liebermann_NG4-94_001.txt',
        'Tavern at the waterfront',
        'M. Tiebermann, Gartenlokal An Der Havel Nikolskoe, 1916',
        'en_US/cmu-arctic_low#ljm',
    ],
    # 12
    [
        '12_Slevogt_AII1022_001.jpg',
        '12_Slevogt_AII1022_001.txt',
        'sailing yachts pool fluss',
        'M. Slevogt, Segelboote Auf Der Alster Am Abend, 1905',
        'en_US/m-ailabs_low#mary_ann',
    ],
]


SILENT_VIDEO = '_silent_video.mp4'


# SILENT CLIP


for img, text, scene, title, voice in DESCRIPTIONS[2:4]:
    
    

    # TITLE BANNER
    
    im = cv2.imread(PIC_DIR + img)  # IMG must have EVEN shape
    h, w, _ = im.shape
    im = im[(h%2):, (w%2):, :]  # assure even image
    print(im.shape, "GLOBAL IM\n\n\n\n")
    fram = np.zeros((94, im.shape[1], 3), dtype=np.uint8)
    h, w, _ = fram.shape
    font                   = cv2.FONT_HERSHEY_SIMPLEX
    bottomLeftCornerOfText = (4, 74)  # OFFSET IN BANNER (240, 74)  # w,h 
    fontScale              = 2
    fontColor              = (255, 255, 255)
    thickness              = 4
    lineType               = 2
    cv2.putText(fram, title,
        bottomLeftCornerOfText,
        font,
        fontScale,
        fontColor,
        thickness,
        lineType)
    offset_h = 24
    im[offset_h:h+offset_h, :w, :] = (.4 * im[offset_h:h+offset_h, :w, :] + .6 * fram).astype(np.uint8)    

    cv2.imwrite('_tmp_banner.png', im)
    print('save 2nd',PIC_DIR + 'thumb____' + img)
    cv2.imwrite(PIC_DIR + 'thumb____' + img, cv2.resize(im, (im.shape[1]//4, im.shape[0]//4)))
    OUT_FILE = img.split('/')[-1].replace('.','__') + '.mp4'  # assets / -1
    print(f'{OUT_FILE=}\n')
    subprocess.run(
            [   
             "python",
             "tts.py",
             "--text", PIC_DIR + text,
             '--image', '_tmp_banner.png',
             '--scene', scene,
             '--voice', voice,
             '--out_file', OUT_FILE,  # save to correct location is handled in client
                ])