diff --git a/doc/filters.texi b/doc/filters.texi
index 1c51e42388..19004f233f 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -1193,6 +1193,10 @@ select gain to noise approach, this is most popular one.
 Set gain to be applied to IR coefficients before filtering.
 Allowed range is 0 to 1. This can be set even with @var{again} used.
 
+@item irfmt
+Set format of IR stream. Can be @code{mono} or @code{input}.
+Default is @code{input}.
+
 @item maxir
 Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds.
 Allowed range is 0.1 to 60 seconds.
diff --git a/libavfilter/af_afir.c b/libavfilter/af_afir.c
index 344905ad5a..b2d158c17e 100644
--- a/libavfilter/af_afir.c
+++ b/libavfilter/af_afir.c
@@ -533,7 +533,7 @@ static int query_formats(AVFilterContext *ctx)
         AV_PIX_FMT_RGB0,
         AV_PIX_FMT_NONE
     };
-    int ret, i;
+    int ret;
 
     if (s->response) {
         AVFilterLink *videolink = ctx->outputs[1];
@@ -543,12 +543,25 @@ static int query_formats(AVFilterContext *ctx)
     }
 
     layouts = ff_all_channel_counts();
-    if ((ret = ff_channel_layouts_ref(layouts, &ctx->outputs[0]->in_channel_layouts)) < 0)
-        return ret;
+    if (!layouts)
+        return AVERROR(ENOMEM);
+
+    if (s->ir_format) {
+        ret = ff_set_common_channel_layouts(ctx, layouts);
+        if (ret < 0)
+            return ret;
+    } else {
+        AVFilterChannelLayouts *mono = NULL;
 
-    for (i = 0; i < 2; i++) {
-        layouts = ff_all_channel_counts();
-        if ((ret = ff_channel_layouts_ref(layouts, &ctx->inputs[i]->out_channel_layouts)) < 0)
+        ret = ff_add_channel_layout(&mono, AV_CH_LAYOUT_MONO);
+        if (ret)
+            return ret;
+
+        if ((ret = ff_channel_layouts_ref(layouts, &ctx->inputs[0]->out_channel_layouts)) < 0)
+            return ret;
+        if ((ret = ff_channel_layouts_ref(layouts, &ctx->outputs[0]->in_channel_layouts)) < 0)
+            return ret;
+        if ((ret = ff_channel_layouts_ref(mono, &ctx->inputs[1]->out_channel_layouts)) < 0)
             return ret;
     }
 
@@ -565,14 +578,6 @@ static int config_output(AVFilterLink *outlink)
     AVFilterContext *ctx = outlink->src;
     AudioFIRContext *s = ctx->priv;
 
-    if (ctx->inputs[0]->channels != ctx->inputs[1]->channels &&
-        ctx->inputs[1]->channels != 1) {
-        av_log(ctx, AV_LOG_ERROR,
-               "Second input must have same number of channels as first input or "
-               "exactly 1 channel.\n");
-        return AVERROR(EINVAL);
-    }
-
     s->one2many = ctx->inputs[1]->channels == 1;
     outlink->sample_rate = ctx->inputs[0]->sample_rate;
     outlink->time_base   = ctx->inputs[0]->time_base;
@@ -739,6 +744,9 @@ static const AVOption afir_options[] = {
     {  "dc",    "DC gain",           0,                  AV_OPT_TYPE_CONST, {.i64=1},    0,  0, AF, "gtype" },
     {  "gn",    "gain to noise",     0,                  AV_OPT_TYPE_CONST, {.i64=2},    0,  0, AF, "gtype" },
     { "irgain", "set IR gain",       OFFSET(ir_gain),    AV_OPT_TYPE_FLOAT, {.dbl=1},    0,  1, AF },
+    { "irfmt",  "set IR format",     OFFSET(ir_format),  AV_OPT_TYPE_INT,   {.i64=1},    0,  1, AF, "irfmt" },
+    {  "mono",  "single channel",    0,                  AV_OPT_TYPE_CONST, {.i64=0},    0,  0, AF, "irfmt" },
+    {  "input", "same as input",     0,                  AV_OPT_TYPE_CONST, {.i64=1},    0,  0, AF, "irfmt" },
     { "maxir",  "set max IR length", OFFSET(max_ir_len), AV_OPT_TYPE_FLOAT, {.dbl=30}, 0.1, 60, AF },
     { "response", "show IR frequency response", OFFSET(response), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, VF },
     { "channel", "set IR channel to display frequency response", OFFSET(ir_channel), AV_OPT_TYPE_INT, {.i64=0}, 0, 1024, VF },
diff --git a/libavfilter/af_afir.h b/libavfilter/af_afir.h
index 9e6f957a8d..ecc0d60641 100644
--- a/libavfilter/af_afir.h
+++ b/libavfilter/af_afir.h
@@ -41,6 +41,7 @@ typedef struct AudioFIRContext {
     int again;
     int gtype;
     float ir_gain;
+    int ir_format;
     float max_ir_len;
     int response;
     int w, h;