transcribe-cpp-sys 0.4.1

Native FFI bindings for transcribe.cpp, a C/C++ speech-to-text library built on ggml
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202

// arch/sortformer/weights.cpp - Sortformer hparam KV reader + weight
// catalog for the Sortformer-specific tensors (encoder_proj + transformer
// + diar head). The conformer weights are loaded separately (parakeet
// reuse); see model.cpp::load.

#include "weights.h"

#include "ggml.h"
#include "gguf.h"
#include "transcribe-log.h"

#include <cstdio>
#include <string>

namespace transcribe::sortformer {

namespace {

transcribe_status kv_u32(const gguf_context * g, const char * key, int32_t & out) {
    const int64_t id = gguf_find_key(g, key);
    if (id < 0) {
        log_msg(TRANSCRIBE_LOG_LEVEL_ERROR, "sortformer: missing KV %s", key);
        return TRANSCRIBE_ERR_GGUF;
    }
    out = static_cast<int32_t>(gguf_get_val_u32(g, id));
    return TRANSCRIBE_OK;
}

transcribe_status kv_f32(const gguf_context * g, const char * key, float & out) {
    const int64_t id = gguf_find_key(g, key);
    if (id < 0) {
        log_msg(TRANSCRIBE_LOG_LEVEL_ERROR, "sortformer: missing KV %s", key);
        return TRANSCRIBE_ERR_GGUF;
    }
    out = gguf_get_val_f32(g, id);
    return TRANSCRIBE_OK;
}

transcribe_status kv_str(const gguf_context * g, const char * key, std::string & out) {
    const int64_t id = gguf_find_key(g, key);
    if (id < 0) {
        log_msg(TRANSCRIBE_LOG_LEVEL_ERROR, "sortformer: missing KV %s", key);
        return TRANSCRIBE_ERR_GGUF;
    }
    out = gguf_get_val_str(g, id);
    return TRANSCRIBE_OK;
}

}  // namespace

transcribe_status read_sortformer_hparams(const gguf_context * g, SortformerHParams & hp) {
    if (g == nullptr) {
        return TRANSCRIBE_ERR_INVALID_ARG;
    }

#define RD_U32(key, field)                                                          \
    if (const transcribe_status st = kv_u32(g, key, hp.field); st != TRANSCRIBE_OK) \
    return st
#define RD_F32(key, field)                                                          \
    if (const transcribe_status st = kv_f32(g, key, hp.field); st != TRANSCRIBE_OK) \
    return st
#define RD_STR(key, field)                                                          \
    if (const transcribe_status st = kv_str(g, key, hp.field); st != TRANSCRIBE_OK) \
    return st

    RD_U32("stt.sortformer.max_speakers", max_speakers);
    RD_U32("stt.sortformer.frame_hop", frame_hop);

    RD_U32("stt.sortformer.encoder.n_layers", enc_n_layers);
    RD_U32("stt.sortformer.encoder.d_model", enc_d_model);
    RD_U32("stt.sortformer.encoder.n_heads", enc_n_heads);
    RD_U32("stt.sortformer.encoder.d_ff", enc_d_ff);
    RD_U32("stt.sortformer.encoder.conv_kernel", enc_conv_kernel);
    RD_U32("stt.sortformer.encoder.subsampling_factor", enc_subsampling_factor);
    RD_U32("stt.sortformer.encoder.subsampling_channels", enc_subsampling_channels);
    RD_U32("stt.sortformer.encoder.feat_in", enc_feat_in);
    RD_U32("stt.sortformer.encoder.pos_emb_max_len", enc_pos_emb_max_len);
    RD_STR("stt.sortformer.encoder.conv_norm_type", enc_conv_norm_type);

    RD_U32("stt.sortformer.transformer.n_layers", tf_n_layers);
    RD_U32("stt.sortformer.transformer.d_model", tf_d_model);
    RD_U32("stt.sortformer.transformer.n_heads", tf_n_heads);
    RD_U32("stt.sortformer.transformer.d_ff", tf_d_ff);
    RD_STR("stt.sortformer.transformer.activation", tf_activation);
    {
        const int64_t id = gguf_find_key(g, "stt.sortformer.transformer.pre_ln");
        hp.tf_pre_ln     = (id >= 0) && gguf_get_val_bool(g, id);
    }

    RD_U32("stt.frontend.num_mels", fe_num_mels);
    RD_U32("stt.frontend.sample_rate", fe_sample_rate);
    RD_U32("stt.frontend.n_fft", fe_n_fft);
    RD_U32("stt.frontend.win_length", fe_win_length);
    RD_U32("stt.frontend.hop_length", fe_hop_length);
    RD_STR("stt.frontend.window", fe_window);
    RD_STR("stt.frontend.normalize", fe_normalize);
    RD_F32("stt.frontend.dither", fe_dither);
    RD_F32("stt.frontend.pre_emphasis", fe_pre_emphasis);

    RD_U32("stt.sortformer.stream.chunk_len", stream_chunk_len);
    RD_U32("stt.sortformer.stream.spkcache_len", stream_spkcache_len);
    RD_U32("stt.sortformer.stream.fifo_len", stream_fifo_len);
    RD_U32("stt.sortformer.stream.spkcache_update_period", stream_spkcache_update_period);

#undef RD_U32
#undef RD_F32
#undef RD_STR

    if (hp.tf_n_heads == 0 || hp.tf_d_model % hp.tf_n_heads != 0) {
        log_msg(TRANSCRIBE_LOG_LEVEL_ERROR, "sortformer: tf_d_model %d not divisible by tf_n_heads %d", hp.tf_d_model,
                hp.tf_n_heads);
        return TRANSCRIBE_ERR_GGUF;
    }
    if (hp.tf_pre_ln) {
        log_msg(TRANSCRIBE_LOG_LEVEL_ERROR, "sortformer: pre_ln=true transformer not supported (post-LN only)");
        return TRANSCRIBE_ERR_GGUF;
    }
    return TRANSCRIBE_OK;
}

namespace {

// Look up a tensor by name and validate its ne against the expected dims
// (fast-to-slow; pass -1 to skip a dim). Returns null + logs on failure.
ggml_tensor * get_checked(ggml_context * ctx, const char * name, int64_t ne0, int64_t ne1) {
    ggml_tensor * t = ggml_get_tensor(ctx, name);
    if (t == nullptr) {
        log_msg(TRANSCRIBE_LOG_LEVEL_ERROR, "sortformer: missing tensor %s", name);
        return nullptr;
    }
    if ((ne0 >= 0 && t->ne[0] != ne0) || (ne1 >= 0 && t->ne[1] != ne1)) {
        log_msg(TRANSCRIBE_LOG_LEVEL_ERROR, "sortformer: tensor %s shape mismatch: have [%lld,%lld] want [%lld,%lld]",
                name, (long long) t->ne[0], (long long) t->ne[1], (long long) ne0, (long long) ne1);
        return nullptr;
    }
    return t;
}

}  // namespace

transcribe_status build_sortformer_weights(ggml_context *            ctx,
                                           const SortformerHParams & hp,
                                           SortformerWeights &       w,
                                           const char *              tensor_prefix) {
    const int64_t d   = hp.tf_d_model;
    const int64_t dff = hp.tf_d_ff;
    const int64_t ed  = hp.enc_d_model;
    const int64_t spk = hp.max_speakers;
    const char *  pfx = (tensor_prefix != nullptr) ? tensor_prefix : "";
    char          name[160];

#define GET(dst, nm, e0, e1)                                  \
    do {                                                      \
        std::snprintf(name, sizeof(name), "%s%s", pfx, (nm)); \
        (dst) = get_checked(ctx, name, (e0), (e1));           \
        if ((dst) == nullptr)                                 \
            return TRANSCRIBE_ERR_GGUF;                       \
    } while (0)

    GET(w.enc_proj_w, "diar.encoder_proj.weight", ed, d);
    GET(w.enc_proj_b, "diar.encoder_proj.bias", d, -1);

    w.tf_blocks.resize(static_cast<size_t>(hp.tf_n_layers));
    for (int i = 0; i < hp.tf_n_layers; ++i) {
        SortformerTfBlock & b = w.tf_blocks[static_cast<size_t>(i)];
        char                blk[96];
#define GETB(dst, suffix, e0, e1)                                      \
    do {                                                               \
        std::snprintf(blk, sizeof(blk), "tf.blocks.%d.%s", i, suffix); \
        GET(dst, blk, e0, e1);                                         \
    } while (0)
        GETB(b.norm1_w, "norm_1.weight", d, -1);
        GETB(b.norm1_b, "norm_1.bias", d, -1);
        GETB(b.attn_q_w, "attn.q.weight", d, d);
        GETB(b.attn_q_b, "attn.q.bias", d, -1);
        GETB(b.attn_k_w, "attn.k.weight", d, d);
        GETB(b.attn_k_b, "attn.k.bias", d, -1);
        GETB(b.attn_v_w, "attn.v.weight", d, d);
        GETB(b.attn_v_b, "attn.v.bias", d, -1);
        GETB(b.attn_o_w, "attn.out.weight", d, d);
        GETB(b.attn_o_b, "attn.out.bias", d, -1);
        GETB(b.norm2_w, "norm_2.weight", d, -1);
        GETB(b.norm2_b, "norm_2.bias", d, -1);
        GETB(b.ff_in_w, "ff.in.weight", d, dff);
        GETB(b.ff_in_b, "ff.in.bias", dff, -1);
        GETB(b.ff_out_w, "ff.out.weight", dff, d);
        GETB(b.ff_out_b, "ff.out.bias", d, -1);
#undef GETB
    }

    GET(w.fc1_w, "diar.fc1.weight", d, d);
    GET(w.fc1_b, "diar.fc1.bias", d, -1);
    GET(w.single_spk_head_w, "diar.single_spk_head.weight", d, spk);
    GET(w.single_spk_head_b, "diar.single_spk_head.bias", spk, -1);
    GET(w.spk_head_w, "diar.spk_head.weight", 2 * d, spk);
    GET(w.spk_head_b, "diar.spk_head.bias", spk, -1);

#undef GET
    return TRANSCRIBE_OK;
}

}  // namespace transcribe::sortformer