Recording starts but doesn't paste to Anki after ending.

[TuxEnthusiast]

I downloaded this script a couple of days ago and it worked perfectly. I don't really know what happened, but now it just gives me the "Recording..." notification and when I hit my bind again nothing happens. Sometimes it does end up adding the card minutes after I've ended the recording. Nothing gets printed when I enter the command into the terminal. I don't know what to do. Please help. I love this tool. Thank you.

It seems to be stuck in this loop:

OS: Fedora Linux 41 (KDE Plasma)KERNEL: 6.11.5-300.fc41.x86_64CPU: Intel Core i7-7700K @ 4.20GHzGPU: NVIDIA GeForce GTX 1070GPU DRIVER: NVIDIA 560.35.03RAM: 32 GB

[stephen-huan]

Duplicate of #14? We haven't actually fixed #14 yet, so feel free to keep the issue open...

[WaddlesPlays]

I ran into this issue today, and using some insight from #14 I was able to get a working config by replacing the recording function with pw-record -P '{ stream.capture.sink=true }' "$audio_file" and removing a few things that didn't seem to be needed anymore from record_start and record_end. It does rely on pipewire, and may not conform to any given setup, but it should be a solid start!

Here is the config:

record_function() {
    local -r audio_file="$1"
    pw-record -P '{ stream.capture.sink=true }' "$audio_file" 1> /dev/null &
}

record_start() {
    # begin recording audio.
    local -r audio_file="$(mktemp \
                               "/tmp/pipewire-recording.XXXXXX.$AUDIO_FORMAT")"
    echo "$audio_file" > "$recording_toggle"

    record_function "$audio_file"
    echo "$!" >> "$recording_toggle"

    current_time >> "$recording_toggle"

    notify_record_start
}

record_end() {
    # end recording.
    local -r audio_file="$(sed -n "1p" "$recording_toggle")"
    local -r pid="$(sed -n "2p" "$recording_toggle")"
    local -r start_time="$(sed -n "3p" "$recording_toggle")"
    local -r duration="$(($(current_time) - start_time))"

    if [ "$duration" -le "$MINIMUM_DURATION" ]; then
        sleep "$((MINIMUM_DURATION - duration))e-3"
    fi

    rm "$recording_toggle"
    kill -15 "$pid"

    while [ "$(du "$audio_file" | awk '{ print $1 }')" -eq 0 ]; do
        true
    done

    store_file "${audio_file}"
    update_sound "$(basename -- "$audio_file")"

    notify_record_stop
}

[TuxEnthusiast]

I ran into this issue today, and using some insight from #14 I was able to get a working config by replacing the recording function with pw-record -P '{ stream.capture.sink=true }' "$audio_file" and removing a few things that didn't seem to be needed anymore from record_start and record_end. It does rely on pipewire, and may not conform to any given setup, but it should be a solid start!
Here is the config:

This actually fixed it for me. Thank you!

[WilsonNet]

Oh we have very different setups and you also had the same problem as me, maybe it is an ffmpeg or kernel issue?

Anyway, my setup here if anyone is interested to dive further on the issue.

OS: Arch Linux x86_64
Host: Redmi Book Pro 14 2022
Kernel: Linux 6.11.5-arch1-1
CPU: AMD Ryzen 7 6800H (16) @ 4.79 GHz
GPU: AMD Radeon 680M [Integrated]
Memory: 8.93 GiB / 14.82 GiB (60%)
Window Manager: AWM (xorg)
Pipewire

[WilsonNet]

My version with the pw-record command (feel free to create a PR with this if anybody wants, I'm too lazy for that

record_function() {
    local -r audio_file="$1"
    # We need the & at the end otherwise it hangs in the pw-record
    pw-record -P '{ stream.capture.sink=true }' "$audio_file" &
}

record_start() {
    # begin recording audio.
    local -r audio_file="$(mktemp \
                               "/tmp/ffmpeg-recording.XXXXXX.$AUDIO_FORMAT")"
    echo "$audio_file" >"$recording_toggle"

    record_function "$audio_file"
    echo "$!" >> "$recording_toggle"

    current_time >> "$recording_toggle"

    notify_record_start
}

record_end() {
    local -r audio_file="$(sed -n "1p" "$recording_toggle")"
    local -r pid="$(sed -n "2p" "$recording_toggle")"
    local -r start_time="$(sed -n "3p" "$recording_toggle")"
    local -r duration="$(($(current_time) - start_time))"

    echo "The pid is: ${pid}"
    if [ "$duration" -le "$MINIMUM_DURATION" ]; then
        sleep "$((MINIMUM_DURATION - duration))e-3"
    fi

    rm "$recording_toggle"
    kill -15 "$pid"

    wait "$pid" || true

    while [ "$(du "$audio_file" | awk '{ print $1 }')" -eq 0 ]; do
        true
    done

    store_file "${audio_file}"
    update_sound "$(basename -- "$audio_file")"

    notify_record_stop
}

record() {
    # this section is a heavily modified version of the linux audio
    # script written by salamander on qm's animecards.
    recording_toggle="/tmp/ffmpeg-recording-audio"

    if [[ ! -f /tmp/ffmpeg-recording-audio ]]; then
        record_start
    else
        record_end
    fi
}

[leto-ux]

I use a lot of virtual sinks in my setup, so had to modify the pw-record part a bit. For some reason, just getting the serial, or the object id (which btw, were the same in my case) and putitng them in the --target argument didn't work at all, and using the argument you guys use, that is -P '{ stream.capture.sink=true }', made the entire setup default to one of the virtual sinks that uses the media role Music. So instead i set the target to 0, which makes it not bind to anything and then manually route it via qpwgraph, not very reliable but at least works.

Side note, anyone have issues with the end recording being very quiet? Like way quieter than the source audio?