diff --git a/configure b/configure
index d49638b381..820f88d2c9 100755
--- a/configure
+++ b/configure
@@ -2992,10 +2992,10 @@ wcmv_decoder_select="inflate_wrapper"
 webp_decoder_select="vp8_decoder exif"
 wmalossless_decoder_select="llauddsp"
 wmapro_decoder_select="sinewin wma_freqs"
-wmav1_decoder_select="mdct sinewin wma_freqs"
-wmav1_encoder_select="mdct sinewin wma_freqs"
-wmav2_decoder_select="mdct sinewin wma_freqs"
-wmav2_encoder_select="mdct sinewin wma_freqs"
+wmav1_decoder_select="sinewin wma_freqs"
+wmav1_encoder_select="sinewin wma_freqs"
+wmav2_decoder_select="sinewin wma_freqs"
+wmav2_encoder_select="sinewin wma_freqs"
 wmavoice_decoder_select="lsp rdft dct sinewin"
 wmv1_decoder_select="msmpeg4dec"
 wmv1_encoder_select="msmpeg4enc"
diff --git a/libavcodec/wma.c b/libavcodec/wma.c
index ddf50d087c..41d16e52f8 100644
--- a/libavcodec/wma.c
+++ b/libavcodec/wma.c
@@ -369,7 +369,7 @@ int ff_wma_end(AVCodecContext *avctx)
     int i;
 
     for (i = 0; i < s->nb_block_sizes; i++)
-        ff_mdct_end(&s->mdct_ctx[i]);
+        av_tx_uninit(&s->mdct_ctx[i]);
 
     if (s->use_exp_vlc)
         ff_free_vlc(&s->exp_vlc);
diff --git a/libavcodec/wma.h b/libavcodec/wma.h
index 80e52687fd..5dc604154d 100644
--- a/libavcodec/wma.h
+++ b/libavcodec/wma.h
@@ -24,9 +24,9 @@
 
 #include "libavutil/float_dsp.h"
 #include "libavutil/mem_internal.h"
+#include "libavutil/tx.h"
 
 #include "avcodec.h"
-#include "fft.h"
 #include "get_bits.h"
 #include "put_bits.h"
 
@@ -115,8 +115,9 @@ typedef struct WMACodecContext {
     float max_exponent[MAX_CHANNELS];
     WMACoef coefs1[MAX_CHANNELS][BLOCK_MAX_SIZE];
     DECLARE_ALIGNED(32, float, coefs)[MAX_CHANNELS][BLOCK_MAX_SIZE];
-    DECLARE_ALIGNED(32, FFTSample, output)[BLOCK_MAX_SIZE * 2];
-    FFTContext mdct_ctx[BLOCK_NB_SIZES];
+    DECLARE_ALIGNED(32, float, output)[BLOCK_MAX_SIZE * 2];
+    AVTXContext *mdct_ctx[BLOCK_NB_SIZES];
+    av_tx_fn mdct_fn[BLOCK_NB_SIZES];
     const float *windows[BLOCK_NB_SIZES];
     /* output buffer for one frame and the last for IMDCT windowing */
     DECLARE_ALIGNED(32, float, frame_out)[MAX_CHANNELS][BLOCK_MAX_SIZE * 2];
diff --git a/libavcodec/wmadec.c b/libavcodec/wmadec.c
index 4bc6a28daf..15d6fb42b2 100644
--- a/libavcodec/wmadec.c
+++ b/libavcodec/wmadec.c
@@ -111,8 +111,9 @@ static av_cold int wma_decode_init(AVCodecContext *avctx)
 
     /* init MDCT */
     for (i = 0; i < s->nb_block_sizes; i++) {
-        ret = ff_mdct_init(&s->mdct_ctx[i], s->frame_len_bits - i + 1,
-                           1, 1.0 / 32768.0);
+        float scale = 1.0 / 32768.0;
+        ret = av_tx_init(&s->mdct_ctx[i], &s->mdct_fn[i], AV_TX_FLOAT_MDCT,
+                         1, 1 << (s->frame_len_bits - i), &scale, AV_TX_FULL_IMDCT);
         if (ret < 0)
             return ret;
     }
@@ -448,7 +449,8 @@ static int wma_decode_block(WMACodecContext *s)
     int coef_nb_bits, total_gain;
     int nb_coefs[MAX_CHANNELS];
     float mdct_norm;
-    FFTContext *mdct;
+    AVTXContext *mdct;
+    av_tx_fn mdct_fn;
 
 #ifdef TRACE
     ff_tlog(s->avctx, "***decode_block: %d:%d\n",
@@ -757,14 +759,15 @@ static int wma_decode_block(WMACodecContext *s)
     }
 
 next:
-    mdct = &s->mdct_ctx[bsize];
+    mdct = s->mdct_ctx[bsize];
+    mdct_fn = s->mdct_fn[bsize];
 
     for (ch = 0; ch < channels; ch++) {
         int n4, index;
 
         n4 = s->block_len / 2;
         if (s->channel_coded[ch])
-            mdct->imdct_calc(mdct, s->output, s->coefs[ch]);
+            mdct_fn(mdct, s->output, s->coefs[ch], sizeof(float));
         else if (!(s->ms_stereo && ch == 1))
             memset(s->output, 0, sizeof(s->output));
 
diff --git a/libavcodec/wmaenc.c b/libavcodec/wmaenc.c
index 90bb6949cb..2c647af13b 100644
--- a/libavcodec/wmaenc.c
+++ b/libavcodec/wmaenc.c
@@ -92,7 +92,9 @@ static av_cold int encode_init(AVCodecContext *avctx)
 
     /* init MDCT */
     for (i = 0; i < s->nb_block_sizes; i++) {
-        ret = ff_mdct_init(&s->mdct_ctx[i], s->frame_len_bits - i + 1, 0, 1.0);
+        float scale = 1.0f;
+        ret = av_tx_init(&s->mdct_ctx[i], &s->mdct_fn[i], AV_TX_FLOAT_MDCT,
+                         0, 1 << (s->frame_len_bits - i), &scale, 0);
         if (ret < 0)
             return ret;
     }
@@ -112,7 +114,8 @@ static int apply_window_and_mdct(AVCodecContext *avctx, const AVFrame *frame)
     const float *const *audio = (const float *const *) frame->extended_data;
     int len            = frame->nb_samples;
     int window_index   = s->frame_len_bits - s->block_len_bits;
-    FFTContext *mdct   = &s->mdct_ctx[window_index];
+    AVTXContext *mdct  = s->mdct_ctx[window_index];
+    av_tx_fn mdct_fn   = s->mdct_fn[window_index];
     int ch;
     const float *win   = s->windows[window_index];
     int window_len     = 1 << s->block_len_bits;
@@ -124,7 +127,7 @@ static int apply_window_and_mdct(AVCodecContext *avctx, const AVFrame *frame)
         s->fdsp->vector_fmul_reverse(&s->output[window_len], s->frame_out[ch],
                                     win, len);
         s->fdsp->vector_fmul(s->frame_out[ch], s->frame_out[ch], win, len);
-        mdct->mdct_calc(mdct, s->coefs[ch], s->output);
+        mdct_fn(mdct, s->coefs[ch], s->output, sizeof(float));
         if (!isfinite(s->coefs[ch][0])) {
             av_log(avctx, AV_LOG_ERROR, "Input contains NaN/+-Inf\n");
             return AVERROR(EINVAL);