diff --git a/.drone.yml b/.drone.yml
index 33803de..0bd5e51 100644
--- a/.drone.yml
+++ b/.drone.yml
@@ -7,9 +7,47 @@ platform:
   arch: ''
 
 steps:
-  - name: Test
+  - name: install dependencies
     image: thetadev256/ucast-dev
+    volumes:
+      - name: cache
+        path: /root/.cache
     commands:
       - poetry install
+
+  - name: lint
+    image: thetadev256/ucast-dev
+    volumes:
+      - name: cache
+        path: /root/.cache
+    commands:
       - poetry run invoke lint
+
+  - name: start worker
+    image: thetadev256/ucast-dev
+    volumes:
+      - name: cache
+        path: /root/.cache
+    environment:
+      UCAST_REDIS_HOST: redis
+    commands:
+      - poetry run invoke worker
+    detach: true
+
+  - name: test
+    image: thetadev256/ucast-dev
+    volumes:
+      - name: cache
+        path: /root/.cache
+    environment:
+      UCAST_REDIS_HOST: redis
+    commands:
       - poetry run invoke test
+
+services:
+  - name: redis
+    image: redis:alpine
+
+volumes:
+  - name: cache
+    temp: { }
diff --git a/.gitignore b/.gitignore
index d4414bf..261201c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -14,11 +14,6 @@ node_modules
 # Jupyter
 .ipynb_checkpoints
 
-# Media files
-*.webm
-*.mp4
-*.mp3
-
 # Application data
 /_run*
 *.sqlite3
diff --git a/pyproject.toml b/pyproject.toml
index 9975146..d2f33af 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -44,6 +44,9 @@ honcho = "^1.1.0"
 requires = ["poetry-core>=1.0.0"]
 build-backend = "poetry.core.masonry.api"
 
+[tool.pytest.ini_options]
+DJANGO_SETTINGS_MODULE = "ucast_project.settings"
+
 [tool.flake8]
 extend-ignore = "E501"
 
diff --git a/tasks.py b/tasks.py
index 16dd469..62c614f 100644
--- a/tasks.py
+++ b/tasks.py
@@ -91,8 +91,8 @@ def get_cover(c, vid=""):
     cv_file = tests.DIR_TESTFILES / "cover" / f"c{ti}_gradient.png"
     cv_blur_file = tests.DIR_TESTFILES / "cover" / f"c{ti}_blur.png"
 
-    tn_file = youtube.download_thumbnail(vinfo, tn_file)
-    util.download_file(channel_metadata.avatar_url, av_file)
+    youtube.download_thumbnail(vinfo, tn_file)
+    util.download_image_file(channel_metadata.avatar_url, av_file)
 
     cover.create_cover_file(
         tn_file, av_file, title, channel_name, cover.COVER_STYLE_GRADIENT, cv_file
diff --git a/ucast/apps.py b/ucast/apps.py
index 41af053..ad4b110 100644
--- a/ucast/apps.py
+++ b/ucast/apps.py
@@ -4,8 +4,3 @@ from django.apps import AppConfig
 class UcastConfig(AppConfig):
     default_auto_field = "django.db.models.BigAutoField"
     name = "ucast"
-
-    def ready(self):
-        from ucast.tasks import download
-
-        download.schedule_update_channels()
diff --git a/ucast/management/__init__.py b/ucast/management/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/ucast/management/commands/__init__.py b/ucast/management/commands/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/ucast/management/commands/rqscheduler.py b/ucast/management/commands/rqscheduler.py
new file mode 100644
index 0000000..6a357ff
--- /dev/null
+++ b/ucast/management/commands/rqscheduler.py
@@ -0,0 +1,11 @@
+from django_rq.management.commands import rqscheduler
+
+from ucast.tasks import schedule
+
+
+class Command(rqscheduler.Command):
+    def handle(self, *args, **kwargs):
+        print("Starting ucast scheduler")
+        schedule.clear_scheduled_jobs()
+        schedule.register_scheduled_jobs()
+        super(Command, self).handle(*args, **kwargs)
diff --git a/ucast/service/cover.py b/ucast/service/cover.py
index a50ef90..ef195b1 100644
--- a/ucast/service/cover.py
+++ b/ucast/service/cover.py
@@ -197,9 +197,9 @@ def _get_baseimage(thumbnail: Image.Image, style: CoverStyle):
         ctn_width = int(COVER_WIDTH / thumbnail.height * thumbnail.width)
         ctn_x_left = int((ctn_width - COVER_WIDTH) / 2)
 
-        ctn = thumbnail.resize((ctn_width, COVER_WIDTH), Image.LANCZOS).filter(
-            ImageFilter.GaussianBlur(20)
-        )
+        ctn = thumbnail.resize(
+            (ctn_width, COVER_WIDTH), Image.Resampling.LANCZOS
+        ).filter(ImageFilter.GaussianBlur(20))
         cover.paste(ctn, (-ctn_x_left, 0))
 
     return cover
@@ -219,9 +219,9 @@ def _resize_thumbnail(thumbnail: Image.Image) -> Image.Image:
     tn_crop_y_top = int((tn_resize_height - tn_height) / 2)
     tn_crop_y_bottom = tn_resize_height - tn_crop_y_top
 
-    return thumbnail.resize((COVER_WIDTH, tn_resize_height), Image.LANCZOS).crop(
-        (0, tn_crop_y_top, COVER_WIDTH, tn_crop_y_bottom)
-    )
+    return thumbnail.resize(
+        (COVER_WIDTH, tn_resize_height), Image.Resampling.LANCZOS
+    ).crop((0, tn_crop_y_top, COVER_WIDTH, tn_crop_y_bottom))
 
 
 def _prepare_text_background(
@@ -357,7 +357,7 @@ def _draw_text_avatar(
     )
 
     if avatar:
-        avt = avatar.resize((avt_size, avt_size), Image.LANCZOS)
+        avt = avatar.resize((avt_size, avt_size), Image.Resampling.LANCZOS)
 
         circle_mask = Image.new("L", (avt_size, avt_size))
         circle_mask_draw = ImageDraw.Draw(circle_mask)
diff --git a/ucast/service/storage.py b/ucast/service/storage.py
index a8e41b7..fbf5206 100644
--- a/ucast/service/storage.py
+++ b/ucast/service/storage.py
@@ -1,32 +1,11 @@
 import os
 from pathlib import Path
-from typing import Tuple
 
-import slugify
 from django.conf import settings
 
 UCAST_DIRNAME = "_ucast"
 
 
-def _get_slug(str_in: str) -> str:
-    return slugify.slugify(str_in, lowercase=False, separator="_")
-
-
-def _get_unique_slug(str_in: str, root_dir: Path, extension="") -> Tuple[Path, str]:
-    original_slug = _get_slug(str_in)
-    slug = original_slug
-    i = 0
-
-    while True:
-        testfile = root_dir / (slug + extension)
-
-        if not testfile.exists():
-            return testfile, slug
-
-        i += 1
-        slug = f"{original_slug}_{i}"
-
-
 class ChannelFolder:
     def __init__(self, dir_root: Path):
         self.dir_root = dir_root
diff --git a/ucast/service/util.py b/ucast/service/util.py
index a918721..6e260ee 100644
--- a/ucast/service/util.py
+++ b/ucast/service/util.py
@@ -1,4 +1,4 @@
-import shutil
+import io
 from pathlib import Path
 
 import requests
@@ -15,33 +15,45 @@ def download_file(url: str, download_path: Path):
     open(download_path, "wb").write(r.content)
 
 
-def download_image_file(url: str, download_path: Path) -> Path:
-    download_file(url, download_path)
-    img = Image.open(download_path)
-    img_ext = img.format.lower()
-    img.close()
+def download_image_file(url: str, download_path: Path):
+    """
+    Download an image and convert it to the type given
+    by the path.
 
+    :param url: Image URL
+    :param download_path: Download path
+    """
+    r = requests.get(url, allow_redirects=True)
+    r.raise_for_status()
+
+    img = Image.open(io.BytesIO(r.content))
+    img_ext = img.format.lower()
     if img_ext == "jpeg":
         img_ext = "jpg"
 
-    new_path = download_path.with_suffix("." + img_ext)
-    shutil.move(download_path, new_path)
-    return new_path
+    if "." + img_ext == download_path.suffix:
+        open(download_path, "wb").write(r.content)
+    else:
+        img.save(download_path)
 
 
 def resize_avatar(original_file: Path, new_file: Path):
     avatar = Image.open(original_file)
     avatar_new_height = int(AVATAR_SM_WIDTH / avatar.width * avatar.height)
-    avatar = avatar.resize((AVATAR_SM_WIDTH, avatar_new_height), Image.LANCZOS)
+    avatar = avatar.resize(
+        (AVATAR_SM_WIDTH, avatar_new_height), Image.Resampling.LANCZOS
+    )
     avatar.save(new_file)
 
 
 def resize_thumbnail(original_file: Path, new_file: Path):
     thumbnail = Image.open(original_file)
     tn_new_height = int(THUMBNAIL_SM_WIDTH / thumbnail.width * thumbnail.height)
-    thumbnail = thumbnail.resize((THUMBNAIL_SM_WIDTH, tn_new_height), Image.LANCZOS)
+    thumbnail = thumbnail.resize(
+        (THUMBNAIL_SM_WIDTH, tn_new_height), Image.Resampling.LANCZOS
+    )
     thumbnail.save(new_file)
 
 
-def get_slug(str_in: str) -> str:
-    return slugify.slugify(str_in, lowercase=False, separator="_")
+def get_slug(text: str) -> str:
+    return slugify.slugify(text, lowercase=False, separator="_")
diff --git a/ucast/service/youtube.py b/ucast/service/youtube.py
index 00b1c69..ef68cce 100644
--- a/ucast/service/youtube.py
+++ b/ucast/service/youtube.py
@@ -92,7 +92,7 @@ class ChannelMetadata:
     avatar_url: str
 
 
-def download_thumbnail(vinfo: VideoDetails, download_path: Path) -> Path:
+def download_thumbnail(vinfo: VideoDetails, download_path: Path):
     """
     Download the thumbnail image of a YouTube video and save it at the given filepath.
     The thumbnail file ending is added to the path.
@@ -108,7 +108,8 @@ def download_thumbnail(vinfo: VideoDetails, download_path: Path) -> Path:
         logging.info(f"downloading thumbnail {url}...")
 
         try:
-            return util.download_image_file(url, download_path)
+            util.download_image_file(url, download_path)
+            return
         except requests.HTTPError:
             logging.warning(f"downloading thumbnail {url} failed")
             pass
@@ -158,20 +159,20 @@ def download_audio(
 
 def tag_audio(audio_path: Path, vinfo: VideoDetails, cover_path: Path):
     title_text = f"{vinfo.published.date().isoformat()} {vinfo.title}"
+    comment = f"https://youtu.be/{vinfo.id}\n\n{vinfo.description}"
 
-    audio = id3.ID3(audio_path)
-    audio["TPE1"] = id3.TPE1(encoding=3, text=vinfo.channel_name)  # Artist
-    audio["TALB"] = id3.TALB(encoding=3, text=vinfo.channel_name)  # Album
-    audio["TIT2"] = id3.TIT2(encoding=3, text=title_text)  # Title
-    audio["TYER"] = id3.TYER(encoding=3, text=str(vinfo.published.year))  # Year
-    audio["TDAT"] = id3.TDAT(encoding=3, text=vinfo.published.strftime("%d%m"))  # Date
-    audio["COMM"] = id3.COMM(encoding=3, text=f"YT-ID: {vinfo.id}")  # Comment
+    tag = id3.ID3(audio_path)
+    tag["TPE1"] = id3.TPE1(encoding=3, text=vinfo.channel_name)  # Artist
+    tag["TALB"] = id3.TALB(encoding=3, text=vinfo.channel_name)  # Album
+    tag["TIT2"] = id3.TIT2(encoding=3, text=title_text)  # Title
+    tag["TDRC"] = id3.TDRC(encoding=3, text=vinfo.published.date().isoformat())  # Date
+    tag["COMM"] = id3.COMM(encoding=3, text=comment)  # Comment
 
     with open(cover_path, "rb") as albumart:
-        audio["APIC"] = id3.APIC(
+        tag["APIC"] = id3.APIC(
             encoding=3, mime="image/png", type=3, desc="Cover", data=albumart.read()
         )
-    audio.save()
+    tag.save()
 
 
 def channel_url_from_id(channel_id: str) -> str:
@@ -231,22 +232,6 @@ def get_channel_metadata(channel_url: str) -> ChannelMetadata:
     return ChannelMetadata(channel_id, name, description, avatar)
 
 
-def download_avatar(avatar_url: str, download_path: Path) -> Path:
-    """
-    Download the avatar image of a channel. The .jpg file ending
-    is added to the path.
-
-    :param avatar_url: Channel avatar URL
-    :param download_path: Download path
-    :return: Path with file ending
-    """
-    logging.info(f"downloading avatar {avatar_url}...")
-
-    download_path = download_path.with_suffix(".jpg")
-    util.download_file(avatar_url, download_path)
-    return download_path
-
-
 def get_channel_videos_from_feed(channel_id: str) -> List[VideoScraped]:
     """
     Return videos of a channel using YouTube's RSS feed. Using the feed is fast,
diff --git a/ucast/tasks/download.py b/ucast/tasks/download.py
index 924947c..0dedb97 100644
--- a/ucast/tasks/download.py
+++ b/ucast/tasks/download.py
@@ -1,8 +1,6 @@
 import os
-from datetime import datetime
 
 import django_rq
-from django.conf import settings
 from django.utils import timezone
 
 from ucast.models import Channel, Video
@@ -21,10 +19,8 @@ def _get_or_create_channel(channel_id: str) -> Channel:
         channel_slug = Channel.get_new_slug(channel_data.name)
         channel_folder = store.get_channel_folder(channel_slug)
 
-        avatar_file = youtube.download_avatar(
-            channel_data.avatar_url, channel_folder.file_avatar
-        )
-        util.resize_avatar(avatar_file, channel_folder.file_avatar_sm)
+        util.download_image_file(channel_data.avatar_url, channel_folder.file_avatar)
+        util.resize_avatar(channel_folder.file_avatar, channel_folder.file_avatar_sm)
 
         channel = Channel(
             id=channel_id,
@@ -81,9 +77,8 @@ def download_video(video: Video):
     details = youtube.download_audio(video.id, audio_file)
 
     # Download/convert thumbnails
-    tn_path = youtube.download_thumbnail(
-        details, channel_folder.get_thumbnail(video.slug)
-    )
+    tn_path = channel_folder.get_thumbnail(video.slug)
+    youtube.download_thumbnail(details, tn_path)
     util.resize_thumbnail(tn_path, channel_folder.get_thumbnail(video.slug, True))
     cover_file = channel_folder.get_cover(video.slug)
     cover.create_cover_file(
@@ -120,15 +115,6 @@ def import_channel(channel_id: str, limit: int = None):
         _load_scraped_video(vid, channel)
 
 
-def schedule_update_channels():
-    django_rq.get_scheduler().schedule(
-        datetime.utcnow(),
-        update_channels,
-        id="schedule_update_channels",
-        interval=settings.YT_UPDATE_INTERVAL,
-    )
-
-
 def update_channels():
     """
     Update all channels from their RSS feeds and download new videos.
diff --git a/ucast/tasks/schedule.py b/ucast/tasks/schedule.py
new file mode 100644
index 0000000..e1fe798
--- /dev/null
+++ b/ucast/tasks/schedule.py
@@ -0,0 +1,27 @@
+import logging
+from datetime import datetime
+
+import django_rq
+from django.conf import settings
+
+from ucast.tasks import download
+
+scheduler = django_rq.get_scheduler()
+log = logging.getLogger(__name__)
+
+
+def clear_scheduled_jobs():
+    """Delete all scheduled jobs to prevent duplicates"""
+    for job in scheduler.get_jobs():
+        log.debug("Deleting scheduled job %s", job)
+        job.delete()
+
+
+def register_scheduled_jobs():
+    """Register all scheduled jobs"""
+    scheduler.schedule(
+        datetime.utcnow(),
+        download.update_channels,
+        id="schedule_update_channels",
+        interval=settings.YT_UPDATE_INTERVAL,
+    )
diff --git a/ucast/tests/__init__.py b/ucast/tests/__init__.py
index 293ff6c..c066208 100644
--- a/ucast/tests/__init__.py
+++ b/ucast/tests/__init__.py
@@ -1,3 +1,3 @@
 from importlib.resources import files
 
-DIR_TESTFILES = files("ucast.tests.testfiles")
+DIR_TESTFILES = files("ucast.tests._testfiles")
diff --git a/ucast/tests/_testfiles/audio/audio1.mp3 b/ucast/tests/_testfiles/audio/audio1.mp3
new file mode 100644
index 0000000..97431c9
Binary files /dev/null and b/ucast/tests/_testfiles/audio/audio1.mp3 differ
diff --git a/ucast/tests/testfiles/avatar/a1.jpg b/ucast/tests/_testfiles/avatar/a1.jpg
similarity index 100%
rename from ucast/tests/testfiles/avatar/a1.jpg
rename to ucast/tests/_testfiles/avatar/a1.jpg
diff --git a/ucast/tests/testfiles/avatar/a2.jpg b/ucast/tests/_testfiles/avatar/a2.jpg
similarity index 100%
rename from ucast/tests/testfiles/avatar/a2.jpg
rename to ucast/tests/_testfiles/avatar/a2.jpg
diff --git a/ucast/tests/testfiles/avatar/a3.jpg b/ucast/tests/_testfiles/avatar/a3.jpg
similarity index 100%
rename from ucast/tests/testfiles/avatar/a3.jpg
rename to ucast/tests/_testfiles/avatar/a3.jpg
diff --git a/ucast/tests/testfiles/cover/blank.png b/ucast/tests/_testfiles/cover/blank.png
similarity index 100%
rename from ucast/tests/testfiles/cover/blank.png
rename to ucast/tests/_testfiles/cover/blank.png
diff --git a/ucast/tests/testfiles/cover/c1_blur.png b/ucast/tests/_testfiles/cover/c1_blur.png
similarity index 100%
rename from ucast/tests/testfiles/cover/c1_blur.png
rename to ucast/tests/_testfiles/cover/c1_blur.png
diff --git a/ucast/tests/testfiles/cover/c1_gradient.png b/ucast/tests/_testfiles/cover/c1_gradient.png
similarity index 100%
rename from ucast/tests/testfiles/cover/c1_gradient.png
rename to ucast/tests/_testfiles/cover/c1_gradient.png
diff --git a/ucast/tests/testfiles/cover/c1_noavatar.png b/ucast/tests/_testfiles/cover/c1_noavatar.png
similarity index 100%
rename from ucast/tests/testfiles/cover/c1_noavatar.png
rename to ucast/tests/_testfiles/cover/c1_noavatar.png
diff --git a/ucast/tests/testfiles/cover/c2_blur.png b/ucast/tests/_testfiles/cover/c2_blur.png
similarity index 100%
rename from ucast/tests/testfiles/cover/c2_blur.png
rename to ucast/tests/_testfiles/cover/c2_blur.png
diff --git a/ucast/tests/testfiles/cover/c2_gradient.png b/ucast/tests/_testfiles/cover/c2_gradient.png
similarity index 100%
rename from ucast/tests/testfiles/cover/c2_gradient.png
rename to ucast/tests/_testfiles/cover/c2_gradient.png
diff --git a/ucast/tests/testfiles/cover/c3_blur.png b/ucast/tests/_testfiles/cover/c3_blur.png
similarity index 100%
rename from ucast/tests/testfiles/cover/c3_blur.png
rename to ucast/tests/_testfiles/cover/c3_blur.png
diff --git a/ucast/tests/testfiles/cover/c3_gradient.png b/ucast/tests/_testfiles/cover/c3_gradient.png
similarity index 100%
rename from ucast/tests/testfiles/cover/c3_gradient.png
rename to ucast/tests/_testfiles/cover/c3_gradient.png
diff --git a/ucast/tests/testfiles/sources.md b/ucast/tests/_testfiles/sources.md
similarity index 72%
rename from ucast/tests/testfiles/sources.md
rename to ucast/tests/_testfiles/sources.md
index 92f6d05..8fdcc6a 100644
--- a/ucast/tests/testfiles/sources.md
+++ b/ucast/tests/_testfiles/sources.md
@@ -1,10 +1,11 @@
-### Quellen der Thumbnails/Avatarbilder zum Testen
+### Quellen der Thumbnails/Avatarbilder/Audiodateien zum Testen
 
 - a1/t1: [ThetaDev @ Embedded World 2019](https://www.youtube.com/watch?v=ZPxEr4YdWt8), by [ThetaDev](https://www.youtube.com/channel/UCGiJh0NZ52wRhYKYnuZI08Q) (CC-BY)
 - a2/t2: [Sintel - Open Movie by Blender Foundation](https://www.youtube.com/watch?v=eRsGyueVLvQ), by [Blender](https://www.youtube.com/c/BlenderFoundation) (CC-BY)
 - a3/t3: [Systemabsturz Teaser zur DiVOC bb3](https://www.youtube.com/watch?v=uFqgQ35wyYY), by [media.ccc.de](https://www.youtube.com/channel/UC2TXq_t06Hjdr2g_KdKpHQg) (CC-BY)
+- audio1: [No copyright intro free fire intro](https://www.youtube.com/watch?v=I0RRENheeTo), by [Shahzaib Hassan](https://www.youtube.com/channel/UCmLTTbctUZobNQrr8RtX8uQ), (CC-BY)
 
 ### Weitere Testvideos
 
 - [Persuasion (Instrumental) – RYYZN (No Copyright Music)](https://www.youtube.com/watch?v=DWjFW7Yq1fA), by [RYYZN](https://soundcloud.com/ryyzn) (CC-BY)
-- [Small pink flowers | #shorts | Free Stock Video](https://www.youtube.com/watch?v=lcQZ6YwQHiw), by [Shahzaib Hassan](https://www.youtube.com/channel/UCmLTTbctUZobNQrr8RtX8uQ), (CC-BY)
\ No newline at end of file
+- [Small pink flowers | #shorts | Free Stock Video](https://www.youtube.com/watch?v=lcQZ6YwQHiw), by [Shahzaib Hassan](https://www.youtube.com/channel/UCmLTTbctUZobNQrr8RtX8uQ), (CC-BY)
diff --git a/ucast/tests/testfiles/thumbnail/t1.webp b/ucast/tests/_testfiles/thumbnail/t1.webp
similarity index 100%
rename from ucast/tests/testfiles/thumbnail/t1.webp
rename to ucast/tests/_testfiles/thumbnail/t1.webp
diff --git a/ucast/tests/testfiles/thumbnail/t2.webp b/ucast/tests/_testfiles/thumbnail/t2.webp
similarity index 100%
rename from ucast/tests/testfiles/thumbnail/t2.webp
rename to ucast/tests/_testfiles/thumbnail/t2.webp
diff --git a/ucast/tests/testfiles/thumbnail/t3.webp b/ucast/tests/_testfiles/thumbnail/t3.webp
similarity index 100%
rename from ucast/tests/testfiles/thumbnail/t3.webp
rename to ucast/tests/_testfiles/thumbnail/t3.webp
diff --git a/ucast/tests/service/__init__.py b/ucast/tests/service/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/ucast/tests/test_cover.py b/ucast/tests/service/test_cover.py
similarity index 100%
rename from ucast/tests/test_cover.py
rename to ucast/tests/service/test_cover.py
diff --git a/ucast/tests/service/test_storage.py b/ucast/tests/service/test_storage.py
new file mode 100644
index 0000000..3c5a29c
--- /dev/null
+++ b/ucast/tests/service/test_storage.py
@@ -0,0 +1,56 @@
+import os
+import tempfile
+from pathlib import Path
+
+from ucast.service import storage
+
+
+def test_create_channel_folders(settings):
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    settings.DOWNLOAD_ROOT = tmpdir
+
+    store = storage.Storage()
+    cf1 = store.get_channel_folder("ThetaDev")
+    cf2 = store.get_channel_folder("Jeff_Geerling")
+    cf1b = store.get_channel_folder("ThetaDev")
+
+    cf1_path = tmpdir / "ThetaDev"
+    cf2_path = tmpdir / "Jeff_Geerling"
+
+    assert cf1.dir_root == cf1_path
+    assert cf1b.dir_root == cf1_path
+    assert cf2.dir_root == cf2_path
+
+    assert os.path.isdir(cf1_path)
+    assert os.path.isdir(cf2_path)
+
+
+def test_channel_folder():
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    ucast_dir = tmpdir / "_ucast"
+
+    cf = storage.ChannelFolder(tmpdir)
+
+    # Verify internal paths
+    assert cf.file_avatar == ucast_dir / "avatar.jpg"
+    assert cf.file_avatar_sm == ucast_dir / "avatar_sm.webp"
+    assert cf.dir_covers == ucast_dir / "covers"
+    assert cf.dir_thumbnails == ucast_dir / "thumbnails"
+
+    # Create the folder
+    assert not cf.does_exist()
+    cf.create()
+    assert cf.does_exist()
+
+    assert cf.get_cover("my_video_title") == ucast_dir / "covers" / "my_video_title.png"
+    assert (
+        cf.get_thumbnail("my_video_title")
+        == ucast_dir / "thumbnails" / "my_video_title.webp"
+    )
+    assert (
+        cf.get_thumbnail("my_video_title", True)
+        == ucast_dir / "thumbnails" / "my_video_title_sm.webp"
+    )
+    assert cf.get_audio("my_video_title") == tmpdir / "my_video_title.mp3"
diff --git a/ucast/tests/service/test_util.py b/ucast/tests/service/test_util.py
new file mode 100644
index 0000000..f095804
--- /dev/null
+++ b/ucast/tests/service/test_util.py
@@ -0,0 +1,92 @@
+import tempfile
+from pathlib import Path
+
+import pytest
+from PIL import Image, ImageChops
+
+from ucast import tests
+from ucast.service import util
+
+TEST_FILE_URL = "https://yt3.ggpht.com/ytc/AKedOLSnFfmpibLLoqyaYdsF6bJ-zaLPzomII__FrJve1w=s900-c-k-c0x00ffffff-no-rj"
+
+
+def test_download_file():
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    download_file = tmpdir / "download.jpg"
+    expected_tn_file = tests.DIR_TESTFILES / "avatar" / "a1.jpg"
+
+    util.download_file(TEST_FILE_URL, download_file)
+
+    downloaded_avatar = Image.open(download_file)
+    expected_avatar = Image.open(expected_tn_file)
+
+    diff = ImageChops.difference(downloaded_avatar, expected_avatar)
+    assert diff.getbbox() is None
+
+
+def test_download_image_file():
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    download_file = tmpdir / "download.jpg"
+    expected_tn_file = tests.DIR_TESTFILES / "avatar" / "a1.jpg"
+
+    util.download_image_file(TEST_FILE_URL, download_file)
+
+    downloaded_avatar = Image.open(download_file)
+    expected_avatar = Image.open(expected_tn_file)
+
+    diff = ImageChops.difference(downloaded_avatar, expected_avatar)
+    assert diff.getbbox() is None
+
+
+def test_download_image_file_conv():
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    download_file = tmpdir / "download.png"
+    expected_tn_file = tests.DIR_TESTFILES / "avatar" / "a1.jpg"
+
+    util.download_image_file(TEST_FILE_URL, download_file)
+
+    downloaded_avatar = Image.open(download_file)
+    expected_avatar = Image.open(expected_tn_file)
+
+    diff = ImageChops.difference(downloaded_avatar, expected_avatar)
+    assert diff.getbbox() is None
+
+
+def test_resize_avatar():
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    source_file = tests.DIR_TESTFILES / "avatar" / "a1.jpg"
+    resized_file = tmpdir / "avatar.webp"
+
+    util.resize_avatar(source_file, resized_file)
+
+    resized_avatar = Image.open(resized_file)
+    assert resized_avatar.size == (100, 100)
+
+
+def test_resize_thumbnail():
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    source_file = tests.DIR_TESTFILES / "thumbnail" / "t1.webp"
+    resized_file = tmpdir / "thumbnail.webp"
+
+    util.resize_thumbnail(source_file, resized_file)
+
+    resized_thumbnail = Image.open(resized_file)
+    assert resized_thumbnail.size == (360, 202)
+
+
+@pytest.mark.parametrize(
+    "text,expected_slug",
+    [
+        ("Hello World 👋", "Hello_World"),
+        ("ÄäÖöÜüß", "AaOoUuss"),
+        ("오징어 게임", "ojingeo_geim"),
+    ],
+)
+def test_slug(text: str, expected_slug: str):
+    slug = util.get_slug(text)
+    assert slug == expected_slug
diff --git a/ucast/tests/service/test_youtube.py b/ucast/tests/service/test_youtube.py
new file mode 100644
index 0000000..0b45228
--- /dev/null
+++ b/ucast/tests/service/test_youtube.py
@@ -0,0 +1,214 @@
+import datetime
+import io
+import re
+import shutil
+import subprocess
+import tempfile
+from pathlib import Path
+
+import pytest
+from mutagen import id3
+from PIL import Image, ImageChops
+
+from ucast import tests
+from ucast.service import youtube
+
+VIDEO_ID_THETADEV = "ZPxEr4YdWt8"
+VIDEO_ID_SHORT = "lcQZ6YwQHiw"
+VIDEO_ID_PERSUASION = "DWjFW7Yq1fA"
+
+CHANNEL_ID_THETADEV = "UCGiJh0NZ52wRhYKYnuZI08Q"
+CHANNEL_ID_BLENDER = "UCSMOQeBJ2RAnuFungnQOxLg"
+CHANNEL_URL_BLENDER = "https://www.youtube.com/c/BlenderFoundation"
+
+
+@pytest.fixture(scope="module")
+def video_details() -> youtube.VideoDetails:
+    return youtube.get_video_details(VIDEO_ID_THETADEV)
+
+
+def test_download_thumbnail(video_details):
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    tn_file = tmpdir / "thumbnail.webp"
+    expected_tn_file = tests.DIR_TESTFILES / "thumbnail" / "t1.webp"
+
+    youtube.download_thumbnail(video_details, tn_file)
+
+    tn = Image.open(tn_file)
+    expected_tn = Image.open(expected_tn_file)
+
+    diff = ImageChops.difference(tn, expected_tn)
+    assert diff.getbbox() is None
+
+
+def test_get_video_details(video_details):
+    assert video_details.id == VIDEO_ID_THETADEV
+    assert video_details.title == "ThetaDev @ Embedded World 2019"
+    assert video_details.channel_id == "UCGiJh0NZ52wRhYKYnuZI08Q"
+    assert (
+        video_details.description
+        == """This february I spent one day at the Embedded World in Nuremberg. They showed tons of interesting electronics stuff, so I had to take some pictures and videos for you to see ;-)
+
+Sorry for the late upload, I just didn't have time to edit my footage.
+
+Embedded World: https://www.embedded-world.de/
+
+My website: https://thdev.org
+Twitter: https://twitter.com/Theta_Dev"""
+    )
+    assert video_details.duration == 267
+    assert not video_details.is_currently_live
+    assert not video_details.is_livestream
+    assert not video_details.is_short
+    assert video_details.published == datetime.datetime(
+        2019, 6, 2, tzinfo=datetime.timezone.utc
+    )
+
+
+def test_get_video_details_short():
+    vinfo = youtube.get_video_details(VIDEO_ID_SHORT)
+    assert vinfo.id == VIDEO_ID_SHORT
+    assert (
+        vinfo.title
+        == "Small pink flowers | #shorts | Free Stock Video | \
+creative commons short videos | creative #short"
+    )
+    assert not vinfo.is_currently_live
+    assert not vinfo.is_livestream
+    assert vinfo.is_short
+
+
+def test_download_audio():
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    download_file = tmpdir / "download.mp3"
+
+    vinfo = youtube.download_audio(VIDEO_ID_PERSUASION, download_file)
+    assert vinfo.id == VIDEO_ID_PERSUASION
+    assert vinfo.title == "Persuasion (Instrumental) – RYYZN (No Copyright Music)"
+    assert vinfo.duration == 100
+
+    # Check with ffmpeg if the audio file is valid
+    res = subprocess.run(
+        ["ffmpeg", "-i", str(download_file)],
+        capture_output=True,
+        universal_newlines=True,
+    )
+    assert "Stream #0:0: Audio: mp3" in res.stderr
+
+    match = re.search(r"Duration: (\d{2}:\d{2}:\d{2})", res.stderr)
+    assert match[1] == "00:01:40"
+
+
+def test_tag_audio(video_details):
+    tmpdir_o = tempfile.TemporaryDirectory()
+    tmpdir = Path(tmpdir_o.name)
+    audio_file = tmpdir / "audio.mp3"
+    cover_file = tests.DIR_TESTFILES / "cover" / "c1_blur.png"
+    shutil.copyfile(tests.DIR_TESTFILES / "audio" / "audio1.mp3", audio_file)
+
+    youtube.tag_audio(audio_file, video_details, cover_file)
+
+    tag = id3.ID3(audio_file)
+    assert tag["TPE1"].text[0] == "ThetaDev"
+    assert tag["TALB"].text[0] == "ThetaDev"
+    assert tag["TIT2"].text[0] == "2019-06-02 ThetaDev @ Embedded World 2019"
+    assert tag["TDRC"].text[0].text == "2019-06-02"
+    assert (
+        tag["COMM::XXX"].text[0]
+        == """https://youtu.be/ZPxEr4YdWt8
+
+This february I spent one day at the Embedded World in Nuremberg. They showed tons of interesting electronics stuff, so I had to take some pictures and videos for you to see ;-)
+
+Sorry for the late upload, I just didn't have time to edit my footage.
+
+Embedded World: https://www.embedded-world.de/
+
+My website: https://thdev.org
+Twitter: https://twitter.com/Theta_Dev"""
+    )
+
+    tag_cover = tag["APIC:Cover"]
+    assert tag_cover.mime == "image/png"
+
+    tag_cover_img = Image.open(io.BytesIO(tag_cover.data))
+    expected_cover_img = Image.open(cover_file)
+    diff = ImageChops.difference(tag_cover_img, expected_cover_img)
+    assert diff.getbbox() is None
+
+
+@pytest.mark.parametrize(
+    "channel_str,channel_url",
+    [
+        (
+            "https://www.youtube.com/channel/UCGiJh0NZ52wRhYKYnuZI08Q",
+            "https://www.youtube.com/channel/UCGiJh0NZ52wRhYKYnuZI08Q",
+        ),
+        (
+            "https://www.youtube.com/c/MrBeast6000",
+            "https://www.youtube.com/c/MrBeast6000",
+        ),
+        (
+            "https://www.youtube.com/user/LinusTechTips",
+            "https://www.youtube.com/user/LinusTechTips",
+        ),
+        (
+            "UCGiJh0NZ52wRhYKYnuZI08Q",
+            "https://www.youtube.com/channel/UCGiJh0NZ52wRhYKYnuZI08Q",
+        ),
+        (
+            "https://piped.mha.fi/user/LinusTechTips",
+            "https://www.youtube.com/user/LinusTechTips",
+        ),
+    ],
+)
+def test_channel_url_from_str(channel_str: str, channel_url: str):
+    url = youtube.channel_url_from_str(channel_str)
+    assert url == channel_url
+
+
+@pytest.mark.parametrize(
+    "channel_url,channel_id,name,avatar_url",
+    [
+        (
+            youtube.channel_url_from_id(CHANNEL_ID_THETADEV),
+            CHANNEL_ID_THETADEV,
+            "ThetaDev",
+            "https://yt3.ggpht.com/ytc/AKedOLSnFfmpibLLoqyaYdsF6bJ-zaLPzomII__FrJve1w=s900-c-k-c0x00ffffff-no-rj",
+        ),
+        (
+            CHANNEL_URL_BLENDER,
+            CHANNEL_ID_BLENDER,
+            "Blender",
+            "https://yt3.ggpht.com/ytc/AKedOLT_31fFSD3FWEBnHZnyZeJx-GPHJwYCQKcEpaq8NQ=s900-c-k-c0x00ffffff-no-rj",
+        ),
+    ],
+)
+def test_channel_metadata(
+    channel_url: str, channel_id: str, name: str, avatar_url: str
+):
+    metadata = youtube.get_channel_metadata(channel_url)
+    assert metadata.id == channel_id
+    assert metadata.name == name
+    assert metadata.avatar_url == avatar_url
+    assert metadata.description
+
+
+def test_get_channel_videos_from_feed():
+    videos = youtube.get_channel_videos_from_feed(CHANNEL_ID_THETADEV)
+    assert videos
+
+    v1 = videos[0]
+    assert len(v1.id) == 11
+    assert v1.published.tzinfo == datetime.timezone.utc
+    assert v1.published.second > 0 or v1.published.minute > 0 or v1.published.hour > 0
+
+
+def test_get_channel_videos_from_scraper():
+    videos = youtube.get_channel_videos_from_scraper(CHANNEL_ID_THETADEV)
+    assert videos
+
+    v1 = videos[0]
+    assert len(v1.id) == 11
+    assert v1.published is None
diff --git a/ucast/tests/test_util.py b/ucast/tests/test_util.py
deleted file mode 100644
index eb10259..0000000
--- a/ucast/tests/test_util.py
+++ /dev/null
@@ -1,24 +0,0 @@
-import tempfile
-from pathlib import Path
-
-from PIL import Image, ImageChops
-
-from ucast import tests
-from ucast.service import util
-
-TEST_FILE_URL = "https://yt3.ggpht.com/ytc/AKedOLSnFfmpibLLoqyaYdsF6bJ-zaLPzomII__FrJve1w=s900-c-k-c0x00ffffff-no-rj"
-
-
-def test_download_file():
-    tmpdir_o = tempfile.TemporaryDirectory()
-    tmpdir = Path(tmpdir_o.name)
-    download_file = tmpdir / "download.jpg"
-    expected_tn_file = tests.DIR_TESTFILES / "avatar" / "a1.jpg"
-
-    util.download_file(TEST_FILE_URL, download_file)
-
-    downloaded_avatar = Image.open(download_file)
-    expected_avatar = Image.open(expected_tn_file)
-
-    diff = ImageChops.difference(downloaded_avatar, expected_avatar)
-    assert diff.getbbox() is None
diff --git a/ucast/tests/test_youtube.py b/ucast/tests/test_youtube.py
deleted file mode 100644
index 0cedf7c..0000000
--- a/ucast/tests/test_youtube.py
+++ /dev/null
@@ -1,126 +0,0 @@
-import datetime
-import re
-import subprocess
-import tempfile
-from pathlib import Path
-
-import pytest
-from PIL import Image, ImageChops
-
-from ucast import tests
-from ucast.service import youtube
-
-VIDEO_ID_THETADEV = "ZPxEr4YdWt8"
-VIDEO_ID_SHORT = "lcQZ6YwQHiw"
-VIDEO_ID_PERSUASION = "DWjFW7Yq1fA"
-
-CHANNEL_ID_THETADEV = "UCGiJh0NZ52wRhYKYnuZI08Q"
-CHANNEL_ID_BLENDER = "UCSMOQeBJ2RAnuFungnQOxLg"
-CHANNEL_URL_BLENDER = "https://www.youtube.com/c/BlenderFoundation"
-
-
-@pytest.fixture(scope="module")
-def video_info() -> youtube.VideoDetails:
-    return youtube.get_video_details(VIDEO_ID_THETADEV)
-
-
-def test_download_thumbnail(video_info):
-    tmpdir_o = tempfile.TemporaryDirectory()
-    tmpdir = Path(tmpdir_o.name)
-    tn_file = tmpdir / "thumbnail"
-    expected_tn_file = tests.DIR_TESTFILES / "thumbnail" / "t1.webp"
-
-    tn_file = youtube.download_thumbnail(video_info, tn_file)
-    assert tn_file.suffix == ".webp"
-
-    tn = Image.open(tn_file)
-    expected_tn = Image.open(expected_tn_file)
-
-    diff = ImageChops.difference(tn, expected_tn)
-    assert diff.getbbox() is None
-
-
-def test_get_video_info(video_info):
-    assert video_info.id == VIDEO_ID_THETADEV
-    assert video_info.title == "ThetaDev @ Embedded World 2019"
-    assert video_info.channel_id == "UCGiJh0NZ52wRhYKYnuZI08Q"
-    assert (
-        video_info.description
-        == """This february I spent one day at the Embedded World in Nuremberg. They showed tons of interesting electronics stuff, so I had to take some pictures and videos for you to see ;-)
-
-Sorry for the late upload, I just didn't have time to edit my footage.
-
-Embedded World: https://www.embedded-world.de/
-
-My website: https://thdev.org
-Twitter: https://twitter.com/Theta_Dev"""
-    )
-    assert video_info.duration == 267
-    assert not video_info.is_currently_live
-    assert not video_info.is_livestream
-    assert not video_info.is_short
-    assert video_info.published == datetime.datetime(
-        2019, 6, 2, tzinfo=datetime.timezone.utc
-    )
-
-
-def test_get_video_info_short():
-    vinfo = youtube.get_video_details(VIDEO_ID_SHORT)
-    assert vinfo.id == VIDEO_ID_SHORT
-    assert (
-        vinfo.title
-        == "Small pink flowers | #shorts | Free Stock Video | \
-creative commons short videos | creative #short"
-    )
-    assert not vinfo.is_currently_live
-    assert not vinfo.is_livestream
-    assert vinfo.is_short
-
-
-def test_download_video():
-    tmpdir_o = tempfile.TemporaryDirectory()
-    tmpdir = Path(tmpdir_o.name)
-    download_file = tmpdir / "download.mp3"
-
-    vinfo = youtube.download_audio(VIDEO_ID_PERSUASION, download_file)
-    assert vinfo.id == VIDEO_ID_PERSUASION
-    assert vinfo.title == "Persuasion (Instrumental) – RYYZN (No Copyright Music)"
-    assert vinfo.duration == 100
-
-    # Check with ffmpeg if the audio file is valid
-    res = subprocess.run(
-        ["ffmpeg", "-i", str(download_file)],
-        capture_output=True,
-        universal_newlines=True,
-    )
-    assert "Stream #0:0: Audio: mp3" in res.stderr
-
-    match = re.search(r"Duration: (\d{2}:\d{2}:\d{2})", res.stderr)
-    assert match[1] == "00:01:40"
-
-
-@pytest.mark.parametrize(
-    "channel_url,channel_id,name,avatar_url",
-    [
-        (
-            youtube.channel_url_from_id(CHANNEL_ID_THETADEV),
-            CHANNEL_ID_THETADEV,
-            "ThetaDev",
-            "https://yt3.ggpht.com/ytc/AKedOLSnFfmpibLLoqyaYdsF6bJ-zaLPzomII__FrJve1w=s900-c-k-c0x00ffffff-no-rj",
-        ),
-        (
-            CHANNEL_URL_BLENDER,
-            CHANNEL_ID_BLENDER,
-            "Blender",
-            "https://yt3.ggpht.com/ytc/AKedOLT_31fFSD3FWEBnHZnyZeJx-GPHJwYCQKcEpaq8NQ=s900-c-k-c0x00ffffff-no-rj",
-        ),
-    ],
-)
-def test_channel_metadata(
-    channel_url: str, channel_id: str, name: str, avatar_url: str
-):
-    metadata = youtube.get_channel_metadata(channel_url)
-    assert metadata.id == channel_id
-    assert metadata.name == name
-    assert metadata.avatar_url == avatar_url
-    assert metadata.description