speexenc.c

语音压缩和解压缩源代码。

   inFile=argv[optind];
   outFile=argv[optind+1];

   /*Initialize Ogg stream struct*/
   srand(time(NULL));
   if (ogg_stream_init(&os, rand())==-1)
   {
      fprintf(stderr,"Error: stream init failed\n");
      exit(1);
   }

   if (strcmp(inFile, "-")==0)
   {
#if defined WIN32 || defined _WIN32
         _setmode(_fileno(stdin), _O_BINARY);
#endif
      fin=stdin;
   }
   else 
   {
      fin = fopen(inFile, "rb");
      if (!fin)
      {
         perror(inFile);
         exit(1);
      }
      close_in=1;
   }

   {
      fread(first_bytes, 1, 12, fin);
      if (strncmp(first_bytes,"RIFF",4)==0 && strncmp(first_bytes,"RIFF",4)==0)
      {
         if (read_wav_header(fin, &rate, &chan, &fmt, &size)==-1)
            exit(1);
         wave_input=1;
         lsb=1; /* CHECK: exists big-endian .wav ?? */
      }
   }

   if (!mode && !rate)
   {
      /* By default, use narrowband/8 kHz */
      mode=&speex_nb_mode;
      rate=8000;
   } else if (mode && rate)
   {
      if (rate>48000)
      {
         fprintf (stderr, "Error: sampling rate too high: %d Hz, try down-sampling\n", rate);
         exit(1);
      } else if (rate>25000)
      {
         if (mode!=&speex_uwb_mode)
         {
            fprintf (stderr, "Warning: Trying to encode in %s at %d Hz. I'll do it but I suggest you try ultra-wideband instead\n", mode->modeName , rate);
         }
      } else if (rate>12500)
      {
         if (mode!=&speex_wb_mode)
         {
            fprintf (stderr, "Warning: Trying to encode in %s at %d Hz. I'll do it but I suggest you try wideband instead\n", mode->modeName , rate);
         }
      } else if (rate>=6000)
      {
         if (mode!=&speex_nb_mode)
         {
            fprintf (stderr, "Warning: Trying to encode in %s at %d Hz. I'll do it but I suggest you try narrowband instead\n", mode->modeName , rate);
         }
      } else {
         fprintf (stderr, "Error: sampling rate too low: %d Hz\n", rate);
         exit(1);
      }
   } else if (!mode)
   {
      if (rate>48000)
      {
         fprintf (stderr, "Error: sampling rate too high: %d Hz, try down-sampling\n", rate);
         exit(1);
      } else if (rate>25000)
      {
         mode=&speex_uwb_mode;
      } else if (rate>12500)
      {
         mode=&speex_wb_mode;
      } else if (rate>=6000)
      {
         mode=&speex_nb_mode;
      } else {
         fprintf (stderr, "Error: Sampling rate too low: %d Hz\n", rate);
         exit(1);
      }
   } else if (!rate)
   {
      if (mode==&speex_nb_mode)
         rate=8000;
      else if (mode==&speex_wb_mode)
         rate=16000;
      else if (mode==&speex_uwb_mode)
         rate=32000;
   }

   if (!quiet)
      if (rate!=8000 && rate!=16000 && rate!=32000)
         fprintf (stderr, "Warning: Speex is only optimized for 8, 16 and 32 kHz. It will still work at %d Hz but your mileage may vary\n", rate); 

   speex_init_header(&header, rate, 1, mode);
   header.frames_per_packet=nframes;
   header.vbr=vbr_enabled;
   header.nb_channels = chan;

   {
      char *st_string="mono";
      if (chan==2)
         st_string="stereo";
      if (!quiet)
         fprintf (stderr, "Encoding %d Hz audio using %s mode (%s)\n", 
               header.rate, mode->modeName, st_string);
   }
   /*fprintf (stderr, "Encoding %d Hz audio at %d bps using %s mode\n", 
     header.rate, mode->bitrate, mode->modeName);*/

   /*Initialize Speex encoder*/
   st = speex_encoder_init(mode);

   if (strcmp(outFile,"-")==0)
   {
#if defined WIN32 || defined _WIN32
      _setmode(_fileno(stdout), _O_BINARY);
#endif
      fout=stdout;
   }
   else 
   {
      fout = fopen(outFile, "wb");
      if (!fout)
      {
         perror(outFile);
         exit(1);
      }
      close_out=1;
   }


   /*Write header*/
   {

      op.packet = (unsigned char *)speex_header_to_packet(&header, (int*)&(op.bytes));
      op.b_o_s = 1;
      op.e_o_s = 0;
      op.granulepos = 0;
      op.packetno = 0;
      ogg_stream_packetin(&os, &op);
      free(op.packet);

      op.packet = (unsigned char *)comments;
      op.bytes = comments_length;
      op.b_o_s = 0;
      op.e_o_s = 0;
      op.granulepos = 0;
      op.packetno = 1;
      ogg_stream_packetin(&os, &op);
      
      while((result = ogg_stream_flush(&os, &og)))
      {
         if(!result) break;
         ret = oe_write_page(&og, fout);
         if(ret != og.header_len + og.body_len)
         {
            fprintf (stderr,"Error: failed writing header to output stream\n");
            exit(1);
         }
         else
            bytes_written += ret;
      }
   }

   free(comments);

   speex_encoder_ctl(st, SPEEX_GET_FRAME_SIZE, &frame_size);
   speex_encoder_ctl(st, SPEEX_SET_COMPLEXITY, &complexity);
   speex_encoder_ctl(st, SPEEX_SET_SAMPLING_RATE, &rate);

   if (quality >= 0)
   {
      if (vbr_enabled)
         speex_encoder_ctl(st, SPEEX_SET_VBR_QUALITY, &vbr_quality);
      else
         speex_encoder_ctl(st, SPEEX_SET_QUALITY, &quality);
   }
   if (bitrate)
   {
      if (quality >= 0 && vbr_enabled)
         fprintf (stderr, "Warning: --bitrate option is overriding --quality\n");
      speex_encoder_ctl(st, SPEEX_SET_BITRATE, &bitrate);
   }
   if (vbr_enabled)
   {
      tmp=1;
      speex_encoder_ctl(st, SPEEX_SET_VBR, &tmp);
   } else if (vad_enabled)
   {
      tmp=1;
      speex_encoder_ctl(st, SPEEX_SET_VAD, &tmp);
   }
   if (dtx_enabled)
      speex_encoder_ctl(st, SPEEX_SET_DTX, &tmp);
   if (dtx_enabled && !(vbr_enabled || abr_enabled || vad_enabled))
   {
      fprintf (stderr, "Warning: --dtx is useless without --vad, --vbr or --abr\n");
   } else if ((vbr_enabled || abr_enabled) && (vad_enabled))
   {
      fprintf (stderr, "Warning: --vad is already implied by --vbr or --abr\n");
   }

   if (abr_enabled)
   {
      speex_encoder_ctl(st, SPEEX_SET_ABR, &abr_enabled);
   }

   if (denoise_enabled || agc_enabled)
   {
      preprocess = speex_preprocess_state_init(frame_size, rate);
      speex_preprocess_ctl(preprocess, SPEEX_PREPROCESS_SET_DENOISE, &denoise_enabled);
      speex_preprocess_ctl(preprocess, SPEEX_PREPROCESS_SET_AGC, &agc_enabled);
   }

   speex_bits_init(&bits);

   if (!wave_input)
   {
      if (read_samples(fin,frame_size,fmt,chan,lsb,input, first_bytes, NULL))
         eos=1;
   } else {
      if (read_samples(fin,frame_size,fmt,chan,lsb,input, NULL, &size))
         eos=1;
   }
   /*Main encoding loop (one frame per iteration)*/
   while (!eos)
   {
      id++;
      /*Encode current frame*/
      if (chan==2)
         speex_encode_stereo_int(input, frame_size, &bits);

      if (preprocess)
         speex_preprocess(preprocess, input, NULL);

      speex_encode_int(st, input, &bits);
      
      if (print_bitrate) {
         int tmp;
         char ch=13;
         speex_encoder_ctl(st, SPEEX_GET_BITRATE, &tmp);
         fputc (ch, stderr);
         cumul_bits += tmp;
         enc_frames += 1;
         if (!quiet)
         {
            if (vad_enabled || vbr_enabled || abr_enabled)
               fprintf (stderr, "Bitrate is use: %d bps  (average %d bps)   ", tmp, (int)(cumul_bits/enc_frames));
            else
               fprintf (stderr, "Bitrate is use: %d bps     ", tmp);
         }
         
      }

      if (wave_input)
      {
         if (read_samples(fin,frame_size,fmt,chan,lsb,input, NULL, &size))
         {
            eos=1;
            op.e_o_s = 1;
         }
      } else {
         if (read_samples(fin,frame_size,fmt,chan,lsb,input, NULL, NULL))
         {
            eos=1;
            op.e_o_s = 1;
         }
      }

      if ((id+1)%nframes!=0)
         continue;
      speex_bits_insert_terminator(&bits);
      nbBytes = speex_bits_write(&bits, cbits, MAX_FRAME_BYTES);
      speex_bits_reset(&bits);
      op.packet = (unsigned char *)cbits;
      op.bytes = nbBytes;
      op.b_o_s = 0;
      if (eos)
         op.e_o_s = 1;
      else
         op.e_o_s = 0;
      op.granulepos = (id+nframes)*frame_size;
      op.packetno = 2+id/nframes;
      ogg_stream_packetin(&os, &op);

      /*Write all new pages (most likely 0 or 1)*/
      while (ogg_stream_pageout(&os,&og))
      {
         ret = oe_write_page(&og, fout);
         if(ret != og.header_len + og.body_len)
         {
            fprintf (stderr,"Error: failed writing header to output stream\n");
            exit(1);
         }
         else
            bytes_written += ret;
      }
   }
   if ((id+1)%nframes!=0)
   {
      while ((id+1)%nframes!=0)
      {
         id++;
         speex_bits_pack(&bits, 15, 5);
      }
      nbBytes = speex_bits_write(&bits, cbits, MAX_FRAME_BYTES);
      op.packet = (unsigned char *)cbits;
      op.bytes = nbBytes;
      op.b_o_s = 0;
      op.e_o_s = 1;
      op.granulepos = (id+nframes)*frame_size;
      op.packetno = 2+id/nframes;
      ogg_stream_packetin(&os, &op);
   }
   /*Flush all pages left to be written*/
   while (ogg_stream_flush(&os, &og))
   {
      ret = oe_write_page(&og, fout);
      if(ret != og.header_len + og.body_len)
      {
         fprintf (stderr,"Error: failed writing header to output stream\n");
         exit(1);
      }
      else
         bytes_written += ret;
   }
   

   speex_encoder_destroy(st);
   speex_bits_destroy(&bits);
   ogg_stream_clear(&os);

   if (close_in)
      fclose(fin);
   if (close_out)
      fclose(fout);
   return 0;
}

/*                 
 Comments will be stored in the Vorbis style.            
 It is describled in the "Structure" section of
    http://www.xiph.org/ogg/vorbis/doc/v-comment.html

The comment header is decoded as follows:
  1) [vendor_length] = read an unsigned integer of 32 bits
  2) [vendor_string] = read a UTF-8 vector as [vendor_length] octets
  3) [user_comment_list_length] = read an unsigned integer of 32 bits
  4) iterate [user_comment_list_length] times {
     5) [length] = read an unsigned integer of 32 bits
     6) this iteration's user comment = read a UTF-8 vector as [length] octets
     }
  7) [framing_bit] = read a single bit as boolean
  8) if ( [framing_bit]  unset or end of packet ) then ERROR
  9) done.

  If you have troubles, please write to [e-mail].
 */

#define readint(buf, base) (((buf[base+3]<<24)&0xff000000)| \
                           ((buf[base+2]<<16)&0xff0000)| \
                           ((buf[base+1]<<8)&0xff00)| \
  	           	    (buf[base]&0xff))
#define writeint(buf, base, val) do{ buf[base+3]=((val)>>24)&0xff; \
                                     buf[base+2]=((val)>>16)&0xff; \
                                     buf[base+1]=((val)>>8)&0xff; \
                                     buf[base]=(val)&0xff; \
                                 }while(0)

void comment_init(char **comments, int* length, char *vendor_string)
{
  int vendor_length=strlen(vendor_string);
  int user_comment_list_length=0;
  int len=4+vendor_length+4;
  char *p=(char*)malloc(len);
  if(p==NULL){
  }
  writeint(p, 0, vendor_length);
  memcpy(p+4, vendor_string, vendor_length);
  writeint(p, 4+vendor_length, user_comment_list_length);
  *length=len;
  *comments=p;
}
void comment_add(char **comments, int* length, char *tag, char *val)
{
  char* p=*comments;
  int vendor_length=readint(p, 0);
  int user_comment_list_length=readint(p, 4+vendor_length);
  int tag_len=(tag?strlen(tag):0);
  int val_len=strlen(val);
  int len=(*length)+4+tag_len+val_len;

  p=(char*)realloc(p, len);
  if(p==NULL){
  }

  writeint(p, *length, tag_len+val_len);      /* length of comment */
  if(tag) memcpy(p+*length+4, tag, tag_len);  /* comment */
  memcpy(p+*length+4+tag_len, val, val_len);  /* comment */
  writeint(p, 4+vendor_length, user_comment_list_length+1);

  *comments=p;
  *length=len;
}
#undef readint
#undef writeint