126 lines
4.3 KiB
HTML
126 lines
4.3 KiB
HTML
|
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
|
||
|
|
<html>
|
||
|
|
<head>
|
||
|
|
<title>Speex Programming</title>
|
||
|
|
|
||
|
|
<meta http-equiv="content-type"
|
||
|
|
content="text/html; charset=ISO-8859-1">
|
||
|
|
</head>
|
||
|
|
<body>
|
||
|
|
|
||
|
|
<div align="center">
|
||
|
|
<h1>Speex Programming</h1>
|
||
|
|
|
||
|
|
<div align="left">
|
||
|
|
<h2>Encoding</h2>
|
||
|
|
In order to encode speech using Speex, you first need to:<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>#include <speex.h></big></pre>
|
||
|
|
</blockquote>
|
||
|
|
You then need to declare a Speex bit-packing struct<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>SpeexBits bits;</big></pre>
|
||
|
|
</blockquote>
|
||
|
|
and a Speex encoder state<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>void *enc_state;</big></pre>
|
||
|
|
</blockquote>
|
||
|
|
The two are initialized by:<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>speex_bits_init(&bits);</big></pre>
|
||
|
|
|
||
|
|
<pre><big>enc_state = speex_encoder_init(&speex_nb_mode);</big></pre>
|
||
|
|
</blockquote>
|
||
|
|
For wideband coding, <i>speex_nb_mode</i> will be replaced by <i>speex_wb_mode</i>
|
||
|
|
. In most cases, you will need to know the frame size used by the mode you
|
||
|
|
are using. You can get that value in the <i>frame_size</i> variable with:<br>
|
||
|
|
<blockquote><big><tt>speex_encoder_ctl(enc_state, SPEEX_GET_FRAME_SIZE, &frame_size);</tt></big><br>
|
||
|
|
</blockquote>
|
||
|
|
For every input frame:<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>speex_bits_reset(&bits);</big></pre>
|
||
|
|
|
||
|
|
<pre><big>speex_encode(enc_state, input_frame, &bits);</big></pre>
|
||
|
|
|
||
|
|
<pre><big>nbBytes = speex_bits_write(&bits, byte_ptr, MAX_NB_BYTES);</big></pre>
|
||
|
|
</blockquote>
|
||
|
|
where <i>input_frame</i> is a <i>(float *)</i> pointing to the beginning
|
||
|
|
of a speech frame, byte_ptr is a <i>(char *)</i> where the encoded frame will
|
||
|
|
be written, <i>MAX_NB_BYTES</i> is the maximum number of bytes that can be
|
||
|
|
written to <i>byte_ptr</i> without causing an overflow and <i>nbBytes</i>
|
||
|
|
is the number of bytes actually written to <i>byte_ptr</i> (the encoded
|
||
|
|
size in bytes).<br>
|
||
|
|
<br>
|
||
|
|
After you're done with the encoding, free all resources with:<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>speex_bits_destroy(&bits);</big></pre>
|
||
|
|
|
||
|
|
<pre><big>speex_encoder_destroy(&enc_state);</big></pre>
|
||
|
|
</blockquote>
|
||
|
|
That's about it for the encoder.<br>
|
||
|
|
|
||
|
|
<h2>Decoding</h2>
|
||
|
|
In order to encode speech using Speex, you first need to:<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>#include <speex.h></big></pre>
|
||
|
|
</blockquote>
|
||
|
|
You then need to declare a Speex bit-packing struct<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>SpeexBits bits;</big></pre>
|
||
|
|
</blockquote>
|
||
|
|
and a Speex encoder state<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>void *dec_state;</big></pre>
|
||
|
|
</blockquote>
|
||
|
|
The two are initialized by:<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>speex_bits_init(&bits);</big></pre>
|
||
|
|
|
||
|
|
<pre><big>dec_state = speex_decoder_init(&speex_nb_mode);</big></pre>
|
||
|
|
</blockquote>
|
||
|
|
For wideband decoding, <i>speex_nb_mode</i> will be replaced by <i>speex_wb_mode</i>
|
||
|
|
. You can get that value in the <i>frame_size</i> variable with:<br>
|
||
|
|
|
||
|
|
<blockquote><big><tt>speex_decoder_ctl(dec_state, SPEEX_GET_FRAME_SIZE, &frame_size);</tt></big><br>
|
||
|
|
</blockquote>
|
||
|
|
There is also a parameter that can be set for the decoder: whether or not
|
||
|
|
to use a perceptual post-filter. This can be set by:<br>
|
||
|
|
<blockquote><big><tt>speex_decoder_ctl(dec_state, SPEEX_SET_PF, &pf);</tt></big><br>
|
||
|
|
</blockquote>
|
||
|
|
where <i>pf</i> is an <i>int</i> that with value 0 to have the post-filter
|
||
|
|
disabled and 1 to have it enabled.<br>
|
||
|
|
<br>
|
||
|
|
For every input frame:<br>
|
||
|
|
|
||
|
|
<blockquote>
|
||
|
|
<pre><big>speex_bits_read_from(&bits, input_bytes, nbBytes);</big></pre>
|
||
|
|
|
||
|
|
<pre><big>speex_decode(st, &bits, output_frame, 0);</big></pre>
|
||
|
|
</blockquote>
|
||
|
|
where <i>input_bytes</i> is a <i>(char *)</i> containing the bit-stream
|
||
|
|
data received for a frame, <i>nbBytes</i> is the size (in bytes) of that
|
||
|
|
bit-stream, and <i>output_frame</i> is a <i>(float *)</i> and points to the
|
||
|
|
area where the decoded speech frame will be written. The last argument indicates
|
||
|
|
whether the frame we'd like to decode was lost. A value of 0 indicates the
|
||
|
|
normal case where bits points to the bit of the current frame. A value of
|
||
|
|
1 indicates that we don't have the bits for the current frame, in which case
|
||
|
|
the bits argument should be the same as the bits for the last correctly received
|
||
|
|
frame. When a frame is lost, the Speex decoder will do its best to "guess"
|
||
|
|
the sorrect signal.<br>
|
||
|
|
<br>
|
||
|
|
</div>
|
||
|
|
</div>
|
||
|
|
|
||
|
|
</body>
|
||
|
|
</html>
|