common : fix audio loading by miniaudio (#2862)

examples/common-whisper.cpp

@@ -76,30 +76,25 @@ bool read_audio_data(const std::string & fname, std::vector<float>& pcmf32, std:
 
 		fprintf(stderr, "%s: read %zu bytes from stdin\n", __func__, audio_data.size());
     }
-    else if (is_wav_buffer(fname)) {
-			if ((result = ma_decoder_init_memory(audio_data.data(), audio_data.size(), &decoder_config, &decoder)) != MA_SUCCESS) {
-				fprintf(stderr, "Error: failed to open audio data from fname buffer (%s)\n", ma_result_description(result));
-
-				return false;
-			}
-    }
-    else if ((result = ma_decoder_init_file(fname.c_str(), &decoder_config, &decoder)) != MA_SUCCESS) {
+    else if (((result = ma_decoder_init_file(fname.c_str(), &decoder_config, &decoder)) != MA_SUCCESS)) {
 #if defined(WHISPER_FFMPEG)
-			if (ffmpeg_decode_audio(fname, audio_data) != 0) {
-				fprintf(stderr, "error: failed to ffmpeg decode '%s'\n", fname.c_str());
+		if (ffmpeg_decode_audio(fname, audio_data) != 0) {
+			fprintf(stderr, "error: failed to ffmpeg decode '%s'\n", fname.c_str());
 
-				return false;
-			}
+			return false;
+		}
 
-			if ((result = ma_decoder_init_memory(audio_data.data(), audio_data.size(), &decoder_config, &decoder)) != MA_SUCCESS) {
-				fprintf(stderr, "error: failed to read audio data as wav (%s)\n", ma_result_description(result));
+		if ((result = ma_decoder_init_memory(audio_data.data(), audio_data.size(), &decoder_config, &decoder)) != MA_SUCCESS) {
+			fprintf(stderr, "error: failed to read audio data as wav (%s)\n", ma_result_description(result));
 
-				return false;
-			}
+			return false;
+		}
 #else
-		fprintf(stderr, "error: failed to open '%s' file (%s)\n", fname.c_str(), ma_result_description(result));
+		if ((result = ma_decoder_init_memory(fname.c_str(), fname.size(), &decoder_config, &decoder)) != MA_SUCCESS) {
+			fprintf(stderr, "error: failed to read audio data as wav (%s)\n", ma_result_description(result));
 
-		return false;
+			return false;
+		}
 #endif
     }
 

examples/common.cpp

@@ -609,21 +609,6 @@ gpt_vocab::id gpt_sample_top_k_top_p_repeat(
 
 }
 
-bool is_wav_buffer(const std::string buf) {
-    // RIFF ref: https://en.wikipedia.org/wiki/Resource_Interchange_File_Format
-    // WAV ref: https://www.mmsp.ece.mcgill.ca/Documents/AudioFormats/WAVE/WAVE.html
-    if (buf.size() < 12 || buf.substr(0, 4) != "RIFF" || buf.substr(8, 4) != "WAVE") {
-        return false;
-    }
-
-    uint32_t chunk_size = *reinterpret_cast<const uint32_t*>(buf.data() + 4);
-    if (chunk_size + 8 != buf.size()) {
-        return false;
-    }
-
-    return true;
-}
-
 void high_pass_filter(std::vector<float> & data, float cutoff, float sample_rate) {
     const float rc = 1.0f / (2.0f * M_PI * cutoff);
     const float dt = 1.0f / sample_rate;

examples/common.h

@@ -134,9 +134,6 @@ gpt_vocab::id gpt_sample_top_k_top_p_repeat(
 // Audio utils
 //
 
-// Check if a buffer is a WAV audio file
-bool is_wav_buffer(const std::string buf);
-
 // Write PCM data into WAV audio file
 class wav_writer {
 private: