diff --git a/doc/base/classes.xml b/doc/base/classes.xml
index d6808f0f34e..8128771b6aa 100644
--- a/doc/base/classes.xml
+++ b/doc/base/classes.xml
@@ -4177,6 +4177,7 @@
 			<description>
 			Set the sample data for a given sample as an array of floats. The length must be equal to the sample lenght or an error will be produced.
 			For this method, a stereo sample is made from two samples. Thus, in case of a stereo sample, the array length must be twice the length returned by [method sample_get_length].
+			Trying to alter a SAMPLE_FORMAT_IMA_ADPCM sample is not supported. It will throw an error to the console, but will not alter the sample data.
 			</description>
 		</method>
 		<method name="sample_set_data">
diff --git a/servers/audio_server.cpp b/servers/audio_server.cpp
index 74f866afb7a..1f0a2e403ac 100644
--- a/servers/audio_server.cpp
+++ b/servers/audio_server.cpp
@@ -48,26 +48,51 @@ AudioServer *AudioServer::get_singleton() {
 
 void AudioServer::sample_set_signed_data(RID p_sample, const DVector<float>& p_buffer) {
 
+	SampleFormat format = sample_get_format(p_sample);
+
+	ERR_EXPLAIN("IMA ADPCM is not supported.");
+	ERR_FAIL_COND(format==SAMPLE_FORMAT_IMA_ADPCM);
+
 	int len = p_buffer.size();
 	ERR_FAIL_COND( len == 0 );
 
 	DVector<uint8_t> data;
-	data.resize(len*2);
-	DVector<uint8_t>::Write w=data.write();
-
-	int16_t *samples = (int16_t*)w.ptr();
-
+	DVector<uint8_t>::Write w;
 	DVector<float>::Read r = p_buffer.read();
 
-	for(int i=0;i<len;i++) {
+	switch(format) {
+		case SAMPLE_FORMAT_PCM8: {
+			data.resize(len);
+			w=data.write();
 
-		float sample = r[i];
-		sample = Math::floor( sample * (1<<16) );
-		if (sample<-32768)
-			sample=-32768;
-		else if (sample>32767)
-			sample=32767;
-		samples[i]=sample;
+			int8_t *samples8 = (int8_t*)w.ptr();
+
+			for(int i=0;i<len;i++) {
+
+				float sample = Math::floor( r[i] * (1<<8) );
+				if (sample<-128)
+					sample=-128;
+				else if (sample>127)
+					sample=127;
+				samples8[i]=sample;
+			}
+		} break;
+		case SAMPLE_FORMAT_PCM16: {
+			data.resize(len*2);
+			w=data.write();
+
+			int16_t *samples16 = (int16_t*)w.ptr();
+
+			for(int i=0;i<len;i++) {
+
+				float sample = Math::floor( r[i] * (1<<16) );
+				if (sample<-32768)
+					sample=-32768;
+				else if (sample>32767)
+					sample=32767;
+				samples16[i]=sample;
+			}
+		} break;
 	}
 
 	w = DVector<uint8_t>::Write();