diff --git a/_service b/_service
deleted file mode 100644
index 8e988d2..0000000
--- a/_service
+++ /dev/null
@@ -1,7 +0,0 @@
-<services>
-  <service name="download_url">
-    <param name="host">www.ffmpeg.org</param>
-    <param name="protocol">https</param>
-    <param name="path">/releases/ffmpeg-5.1.2.tar.xz</param>
-  </service>
-<service name="download_url"><param name="host">www.ffmpeg.org</param><param name="protocol">https</param><param name="path">/releases/ffmpeg-5.1.2.tar.xz.asc</param></service></services>
\ No newline at end of file
diff --git a/_service:download_url:ffmpeg-5.1.2.tar.xz b/_service:download_url:ffmpeg-5.1.2.tar.xz
deleted file mode 100644
index d6cb8b1..0000000
--- a/_service:download_url:ffmpeg-5.1.2.tar.xz
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:619e706d662c8420859832ddc259cd4d4096a48a2ce1eefd052db9e440eef3dc
-size 10004956
diff --git a/_service:download_url:ffmpeg-5.1.2.tar.xz.asc b/_service:download_url:ffmpeg-5.1.2.tar.xz.asc
deleted file mode 100644
index 454e730..0000000
--- a/_service:download_url:ffmpeg-5.1.2.tar.xz.asc
+++ /dev/null
@@ -1,11 +0,0 @@
------BEGIN PGP SIGNATURE-----
-
-iQFMBAABCAA2FiEE/PmG6hXm4pOlZE8QtDIvBNZ2WNgFAmMwK64YHGZmbXBlZy1k
-ZXZlbEBmZm1wZWcub3JnAAoJELQyLwTWdljYPqcH/1NJ+4sBONB93q9ly/Hcq4si
-MfisOgGY+3CqDxHE3VSzMl91vq0vw4Axbbpk2+3AaELfQtDwVh8i8/8uDZXXgVa7
-1ScGEsq13RltI7UXAMRE0teZlFwg/V+ZGeHgN4dhYoMrp9txW5HVvy7CzJ6/hco5
-m6QdEr8U1POHXDlevHvI67tml1M5zDbNtfcmUa0PbRKZblXpmq/jzie41h4w926o
-eUe9j+tal2JM5LtmM9O5DkeblsygeZ0BtyM3KpMaoBykbqEFeqqN93DpUgKCg2GV
-ZqRDJMycz92MXMh/0nXlAWArcxUwnUW3jSSuNvkFbn/uUJECP/sCx1gkwjQRNeg=
-=elRN
------END PGP SIGNATURE-----
diff --git a/ffmpeg-5.changes b/ffmpeg-5.changes
index 28c4b64..01aff29 100644
--- a/ffmpeg-5.changes
+++ b/ffmpeg-5.changes
@@ -2,47 +2,19 @@
 Mon Sep 26 18:44:55 UTC 2022 - C J <c.j@tuta.io>
 
 - Update to release 5.1.2
-  * avcodec/dstdec: Check for overflow in build_filter()
-  * avformat/spdifdec: Use 64bit to compute bit rate
-  * avformat/rpl: Use 64bit for duration computation
-  * avformat/xwma: Use av_rescale() for duration computation
-  * avformat/sdsdec: Use av_rescale() to avoid intermediate overflow in duration calculation
-  * avformat/sbgdec: Check ts_int in genrate_intervals
-  * avformat/sbgdec: clamp end_ts
-  * avformat/rmdec: check tag_size
-  * avformat/nutdec: Check fields
-  * avformat/flvdec: Use 64bit for sum_flv_tag_size
-  * avformat/jacosubdec: Fix overflow in get_shift()
-  * avformat/genh: Check nb_channels for IMA ADPCM
-  * avformat/dxa: avoid bpc overflows
-  * avformat/dhav: Use 64bit seek_back
-  * avformat/cafdec: Check that nb_frasmes fits within 64bit
-  * avformat/asfdec_o: Limit packet offset
-  * avformat/apm: Use 64bit for bit_rate computation
-  * avformat/ape: Check frames size
-  * avformat/icodec: Check nb_pal
-  * avformat/aiffdec: Use 64bit for block_duration use
-  * avformat/aiffdec: Check block_duration
-  * avformat/mxfdec: only probe max run in
-  * avformat/mxfdec: Check run_in is within 65536
-  * avcodec/mjpegdec: Check for unsupported bayer case
-  * avcodec/tta: Check 24bit scaling for overflow
-  * avcodec/exr: Check preview psize
-  * avcodec/tiff: Fix loop detection
-  * libavformat/hls: Free keys
-  * avcodec/fmvc: Move frame allocation to a later stage
-  * avfilter/vf_showinfo: remove backspaces
-  * avcodec/speedhq: Check width
-  * avcodec/bink: disallow odd positioned scaled blocks
-  * avformat/cafenc: derive Opus frame size from the relevant stream parameters
+  * Add more field checks, add checks against overflows,
+    or outright use larger integer types.
   * avformat/dashdec: Fix crash on invalid input/ENOMEM, fix leak
-  * lavc/videotoolbox: do not pass AVCodecContext to decoder output callback
+  * lavc/videotoolbox: do not pass AVCodecContext to decoder
+    output callback
   * lavc/pthread_frame: always transfer stashed hwaccel state
   * avcodec/arm/sbcenc: avoid callee preserved vfp registers
-  * avformat/riffdec: don't unconditionally overwrite WAVEFORMATEXTENSIBLE layout
-  * avfilter/vf_scale: overwrite the width and height expressions with the original values
-  * lavc/pthread_frame: avoid leaving stale hwaccel state in worker threads
-  * avutil/tests/.gitignore: Add channel_layout testtool
+  * avformat/riffdec: don't unconditionally overwrite
+    WAVEFORMATEXTENSIBLE layout
+  * avfilter/vf_scale: overwrite the width and height expressions
+    with the original values
+  * lavc/pthread_frame: avoid leaving stale hwaccel state in
+    worker threads
 
 -------------------------------------------------------------------
 Sun Sep  4 11:45:19 UTC 2022 - Jan Engelhardt <jengelh@inai.de>